Investigation of the microbiota of dairy cow milk in ...eprints.lib.okayama-u.ac.jp/files/public/5/56832/...Milk composition and SCC were determined using a CombiFoss FT+, and milk

Investigation of the microbiota of dairy

cow milk in relation to mastitis

prevention and milk quality control

2019, March

Wu Haoming

Graduate School of

Environmental and Life Science

(Doctor’s Course)

OKAYAMA UNIVERSITY

CONTENTS

FIGURE CONTENTS ................................................................................ I

TABLE CONTENTS ................................................................................ IV

AKNOLAGEMENT ................................................................................. 1

ABSTRACT ............................................................................................ 2

CHAPTER 1 GENERAL INTRODUCTION .................................................. 7

1.1 EFFECTS OF MILK CONTENTS ON THE HEALTH OF DAIRY COWS ................ 7

1.2 PROGRESS AND SHORTCOMINGS OF AMS MILKING MACHINE .................. 8

1.3 BACTERIA IN MILK ........................................................................ 9

1.3.1 Mastitis ............................................................................. 9

1.3.2 Spoilage .......................................................................... 11

REFERENCE ................................................................................. 12

CHAPTER 2 MONITORING THE MICROBIOTA AND NUTRIENT

COMPOSITION IN DAIRY COW MILK COLLECTED FROM A HERD

MANAGED BY AUTOMATIC MILKING SYSTEMS .................................. 19

2.1 INTRODUCTION ..................................................................... 19

2.2 MATERIALS AND METHODS ................................................... 20

2.2.1 Milk samples ................................................................... 20

2.2.2 Preparation of bacterial DNA .......................................... 21

2.2.3 qPCR ............................................................................... 21

2.2.4 Illumina MiSeq sequencing ............................................. 22

2.2.5 Statistical analysis .......................................................... 22

2.3 RESULTS ................................................................................ 23

2.4 DISCUSSION .......................................................................... 28

2.5 CONCLUSION ......................................................................... 33

2.6 REFERENCE ............................................................................ 33

CHAPTER 3 ASSESSMENT OF MILK MICROBIOTA OF DAIRY COWS

MANAGED BY AUTOMATIC MILKING SYSTEMS .................................. 39

3.1 INTRODUCTION ..................................................................... 39


3.2.1 Sample collection ............................................................ 40


3.2.3 Quantitative real-time PCR ............................................. 41



3.3 RESULTS ................................................................................ 43

3.4 DISCUSSION .......................................................................... 48

3.5 CONCLUSION ......................................................................... 54

3.6 REFERENCES .......................................................................... 54

CHAPTER 4 RUMEN FLUID, FECES, MILK, WATER, FEED, AIRBORNE

DUST, AND BEDDING MICROBIOTA IN DAIRY FARMS MANAGED BY

AUTOMATIC MILKING SYSTEMS ......................................................... 59

4.1 INTRODUCTION ..................................................................... 59




4.2.3 Quantitative real-time PCR for total bacteria ................. 61


4.2.5 Bioinformatics and microbiota characterization ............. 62

4.2.6 Statistical analyses ......................................................... 63

4.3 RESULTS ................................................................................ 63

4.4 DISCUSSION .......................................................................... 67

4.5 CONCLUSION ......................................................................... 72

4.6 REFERENCES .......................................................................... 73

CHAPTER 5 MILK COMPOSITION AND MICROBIOTA IN DAIRY FARMS IN

CHINA ................................................................................................ 77

5.1 INTRODUCTION .......................................................................... 77

5.2MATERIALS AND METHODS ............................................................ 78



5.2.3 Quantitative real-time PCR ............................................. 79



5.3 RESULTS ................................................................................... 80

5.4 DISCUSSION .............................................................................. 86

5.5 CONCLUSION ............................................................................. 87

5.6 REFERENCE ............................................................................... 87

CHAPTER 6 MICROBIOTA OF MANUALLY COLLECTED MILK BETWEEN

CHINESE, VIETNAMESE, AND JAPANESE DAIRY FARMS ....................... 90

6.1 INTRODUCTION .......................................................................... 90

6.2MATERIALS AND METHODS ............................................................ 91


6.2.1.1 Collection of Mastitis and Healthy milk ....................................... 91

6.2.1.2 Collection of milk samples from different countries ................... 92




6.3 RESULTS ................................................................................... 94

6.3.1 Experiment 1 ................................................................... 94

6.3.2 Experiment 2 ................................................................... 97

6.4 DISCUSSION ............................................................................ 101

6.5 CONCLUSION ........................................................................... 102

6.6 REFERENCE ............................................................................. 103

I

Figure Contents

Figure 2.1 Seasonal variation of milk composition by one-way ANOVA ________ 23

Figure 2.2 (A) Relation between Milk Indicators (n=201). (B) The somatic cell

count (SCC) in milk has a significant linear positive relationship with the total

bacterial count (TBC) (n=60). __________________________________________ 24

Figure 2.3 alpha diversity index of milk microbiota by one-way ANOVA. _______ 25

Figure 2.4 LEfSe analysis at multiple taxonomic levels of each month data. ____ 27

Figure 2.5 (A) stacked bar chart of phyla proportion in each month (B) LEfSe

analysis at family levels of each month data. ______________________________ 28

Figure 2.6 The correlation network of different milk components and predominant

family(proportion>1%). _______________________________________________ 31

Figure 3.1 Association heatmap between the relative abundance of bacterial

families, sampling procedure, and sampling month for the milk of cows managed

by automatic milking systems. __________________________________________ 46

Figure 3.2 Principal coordinates analyses of (a) bacterial families and (b) KEGG

pathways in the microbiota of manually and automatically collected milk sampled

in May and July from dairy cows managed by automatic milking systems. ______ 47

Figure 3.3 The KEGG pathways encoded in the microbiota of (a) manually

collected and (b) automatically collected milk. Only the KEGG pathways

significantly affected by sampling month are presented. _____________________ 48

Figure 3.4 The KEGG pathways encoded in the microbiota of manually collected

and automatically collected milk. _______________________________________ 49

II

Figure 3.5 The predicted functions based on FAPROTAX database in the

microbiota of manually collected and automatically collected milk. ____________ 53

Figure 4.1 Heatmap showing the relative abundance of major taxa (detected

at >1.0% at least two different samples) in feed, rumen fluid, feces, milk, bedding,

water, and airborne dust microbiota in two dairy farms managed by an automatic

milking system. ______________________________________________________ 65

Figure 4.2 Family level proportions of feed, rumen fluid, feces, milk, bedding,


milking system. ______________________________________________________ 66

Figure 4.3 Canonical analysis of principal coordinates plot characterizing the

microbiota of feed, rumen fluid, feces, milk, bedding, water, and airborne dust in

the two dairy farms managed by an automatic milking system.________________ 68

Figure 4.4 Pie charts of the percentages of inferred sources of milk microbiota in

two dairy farms managed by an automatic milking system. ___________________ 69

Figure 4.5 The five most abundant taxa in rumen fluid, milk, feces, bedding,

airborne dust, water, and feed microbiota in two dairy farms managed by an

automatic milking system. _____________________________________________ 71

Figure 5.1 Farm variation of milk composition(A) and alpha diversity(B) by one-

way ANOVA. _______________________________________________________ 81


at >1.0% at least two different samples) in milk microbiota in five dairy farms

collected from Heilongjiang and Shanxi. _________________________________ 82

Figure 5.3 Principal coordinates analyses of bacterial families in the microbiota of

manually collected milk sampled from different farm in China. _______________ 83

III

Figure 2.5 LEfSe analysis at family levels of Heilongjiang and Shanxi in China.

___________________________________________________________________ 84

Figure 5.5 The correlation network of different milk components and predominant

family(proportion>1%). _______________________________________________ 85

Figure 6.1 Alpha diversity of Bacteria in Milk of Healthy and Mastitis Cows by one-

way ANOVA. _______________________________________________________ 94

Figure 6.2 Phyla level (A) and Family level (B) proportions of Mastitis and healthy

cow’s milk which manually collected in different month managed by an automatic

milking system. ______________________________________________________ 96

Figure 6.3 (A) Heatmap showing the relative abundance of major taxa (detected

at >1.0% at least two different samples) (B) Significant changed family in milk

between healthy and mastitis dairy cows by one-way ANOVA. (C) Principal

coordinates analyses of bacterial families in Mastitis and healthy cow’s milk which

manually collected in different month managed by an automatic milking system. 97

Figure 6.4 Country variation of milk bacterial alpha diversity by one-way ANOVA.

___________________________________________________________________ 98

Figure 6.5 LEfSe analysis at multiple taxonomic levels of each country milk data.

___________________________________________________________________ 99


at >1.0% at least two different samples) in milk microbiota from china, Japan and

Vietnam. __________________________________________________________ 100


manually collected milk sampled from china, Japan and Vietnam. ___________ 101

IV

Table Contents

Table 3.1. Milk yield and composition of dairy cows and total bacterial

population, total sequence reads, and alpha diversity indices of

microbiota of manually and automatically collected milk sampled in May

and July. ................................................................................................... 43

Table 3.2. Relative abundances of bacterial families in the microbiota of

manually and automatically collected milk sampled in May and July

from dairy cows managed by automatic milking systems. ..................... 44

Table 3.3. The KEGG pathways encoded in the microbiota of manually

and automatically collected milk sampled in May and July from dairy

cows managed by automatic milking systems. ........................................ 51

1

AKNOLAGEMENT

I would like to thank my supervisor, Professor Naoki NISHINO, in particular.

From my arrival in Japan to the present six years, Professor Nishino has shown great

concern for my life and research. With his help and guidance, I can smoothly complete

the experiment design, operation and result analysis. And then compile papers for

publication. He guided and witnessed every step of my scientific research.

I exceedingly appreciate my co-supervisors, Assoc. Prof. Takeshi TSURUTA and

Prof. Hiroaki FUNAHASHI, for their academic comments and persistent help in my

research.

I want to thank to Ph.D. Hongyan Han, Ph.D. Xiaomin Ao, Ph.D. Baiyila Wu,

Ph.D. Gaowa Bai, Ph.D. Tang Thuy Minh, Ph.D. Tran Thi Minh Tu, Ph.D. Kuikui Ni

and students who are studying hard in the Animal Nutrition Laboratory for their help,

accompany, and encouragement.

The scholarship from China Scholarship Council financially supported me during

entire doctoral period.

Finally, I express my deep love to my family, thanks for their continued support

and encouragement.

2

Abstract

In the production of milk and dairy products, bacteria in raw milk may closely be

related to health of the cows, quality of the milk, shelf life and flavor of the dairy

products. Contamination of cow's mammary glands by pathogenic bacteria, such as

Esherichia coli, Staphylococcus aureus, and Streptococcus agalactiae, will cause a

large increase in somatic cell count (SCC) in milk and even the occurrence of mastitis.

Mastitis lowers productivity and increases farm costs, including veterinary diagnostics,

medicines, unpaid special feeding, and even risks leading to elimination of the disease’s

dairy cows. In addition, as a source of various dairy products, bacteria in raw milk, such

as Pseudomonas spp. and Acinetobacter spp. can affect flavor and shelf life of the dairy

products by secreting heat-resistant proteases and lipases. Although the milk

disinfection system can effectively eliminate bacteria in milk, heat-resistant proteases

and lipases in milk may continue to be activated and deteriorate quality of the dairy

products. Overall, the more bacteria are present in milk, the greater risk of cow disease

and milk deterioration the dairy farm may have; hence, in the current milk production,

SCC and total bacterial count (TBC) are important indicators to understand cow health

and milk quality. Effective and reasonable control of bacteria in milk can be a key to

prevent mastitis and improve milk quality, and thus it is of great significance to clarify

factors affecting bacteria in raw milk. In this thesis, five experiments were carried out

to examine the effects of season, regions, milking methods, and cowshed environment

on the microbiota of raw milk. Microbiota was assessed by Illumina MiSeq sequencing

and the gene functions were predicted by Phylogenetic Investigation of Communities

by Reconstruction of Unobserved States (PICRUSt).

Seasonal difference in milk composition and milk microbiota

The microbiota, SCC, and composition of milk collected from a herd managed

by automatic milking systems were examined through a 1-year survey. From February

to December 2014 every two months, a total of 201 milk-test day samples were

collected. Milk composition and SCC were determined using a CombiFoss FT+, and

milk microbiota total bacterial population were assessed by MiSeq amplicon

sequencing and real time-PCR respectively. Among 201 samples, 37 samples (18.4%)

3

had SCC of >300×103/mL and 9 samples had >1,000×103/mL. The SCC was positively

correlated with yield, fat content, protein content, and total bacterial count of the milk.

The proportion of Firmicutes appeared to be greater and that of Proteobacteria were

smaller in hot seasons compared with other seasons. Relative abundances of

Pseudomonadaceae, Enterobacteriaceae, Staphylococcaceae, Micrococcaceae, and

Ruminococcaceae were greater in winter (Dec. and Feb.), those of Streptococcaceae,

Microbacteriaceae, and Enterococcaceae were greater in spring (Apr. and Jun.), and

that of Moraxellaceae was greater in autumn (Oct.) than other seasons. In addition, the

abundances of Pseudomonadaceae and Enterobacteriaceae were positively and those

of Staphylococcaceae, Micrococcaceae, Ruminococcaceae, and Moraxellaceae were

negatively related with milk urea nitrogen, suggesting that milk microbiota could be

manipulated by diet and nutritional management.

Microbiota between manually collected and mechanically collected milk

In the previous study, we assessed the microbiota of individual cow’s milk

automatically stored in cooling cups, which became unable because of technical

development with AMS. Although the sample can be regarded as raw milk, the data for

milk microbiota may not be identical to those for udder milk, because of the effect of

cooling. To clarify the differences between manually collected and automatically

collected milk, samples were collected in May and July from three lactating Holstein

cows managed by AMS. Large differences were seen between May- and July-collected

samples in the microbiota of manually collected milk; the total abundances of

Methylobacteriaceae, Sphingomonadaceae, and Staphylococcaceae were

approximately half (45%) in May-collected samples, whereas there was an

overwhelming abundance (>70%) of Enterobacteriaceae in July-collected samples.

The microbiota of automatically collected milk was substantially different;

Moraxellaceae, Pseudomonadaceae, and Enterococcaceae were found as major

families in both May- and July-collected samples, although their proportions were

marginal in manually collected milk. Assessment by PICRUSt demonstrated that gene

functions related with fermentation, diarrhea, gastroenteritis, septicemia, aromatic

compound degradation, and nosocomial and pneumonia were higher in manually

collected than automatically collected milk. Effect of the collection procedure on the

diversity and predicted gene functions of milk microbiota was clear and samples of

4

manually collected milk were shown appropriate.

Characterization of gut, milk, and cowshed environment microbiota

To examine possible contamination source of milk microbiota, rumen fluid,

feces, milk, water, feed (total mixed ration silage), bedding, and airborne dust were

sampled at two dairy farms managed by AMS. The microbiota on each was assessed

by Illumina MiSeq sequencing. The three most prevalent taxa (Aerococcaceae,

Staphylococcaceae, and Ruminococcaceae at farm 1 and Staphylococcaceae,

Lactobacillaceae, and Ruminococcaceae at farm 2) were shared between milk and

airborne dust microbiota. Indeed, Source Tracker indicated that milk microbiota was

related with airborne dust microbiota. Meanwhile, hierarchical clustering and canonical

analysis of principal coordinates demonstrated that the milk microbiota was associated

with the bedding microbiota but clearly separated from feed, rumen fluid, feces, and

water microbiota. Importance of cowshed management for milk quality control and

mastitis prevention was confirmed and emphasized from this study.

A practical survey of composition and microbiota of the milk in dairy farms

in China

From five dairy farms implementing milking parlor and free barn housing in

Heilongjiang and Shanxi provinces, China, milk samples were manually collected from

five cows and their microbiotas were assessed by Illumina MiSeq sequencing. Milk

composition and SCC were determined using CombiFoss FT+, and N-acetyl-β-D-

glycosaminidase (NAGase) activity was determined by a fluorometric method. Milk

microbiota varied across farms and individual cows, and thus differences between

regions were obscure. Enterobacteriaceae was a most predominant family in about one-

third of the milk samples, and Pseudomonadaceae, Staphylococcaceae, Moraxellaceae,

and Lactobacillaceae were detected at high relative abundances in other samples. The

SCC was positively correlated with fat content, protein content, and total bacterial count

in the milk. The abundance of Enterobacteriaceae was positively and those of

Pseudomonadaceae and Lactobacillaceae were negatively related with lactose content,

and that of Moraxellaceae was negatively related with milk urea nitrogen; hence,

appropriate management of protein-energy balance in cows could be regarded as a

5

measure to suppress contamination of Moraxellaceae (Acinetobacter spp.) in the milk.

Unlike the results in experiment 4, cow-to-cow differences were evident even in the

same farm in this survey.

Microbiota of manually collected milk between Chinese, Vietnamese, and

Japanese dairy farms

Using the data for milk microbiota obtained by ourselves and our co-workers,

the similarities and differences between Chinese, Vietnamese, and Japanese practices

were examined. All milk samples were obtained manually, and diets consisting of corn

silage and concentrates, elephant grass and concentrates, and total mixed ration silage

were fed at Chinese, Vietnamese, and Japanese d, respectively. Milking parlor and free

barn housing was implemented at Chinese and Vietnamese farms, and AMS was

operated at Japanese farms. The milk microbiota was assessed by Illumina MiSeq

sequencing. Diversity of the milk microbiota was higher in Japanese than Chinese and

Vietnamese cows; the number of operational taxonomy unit, Chao 1 index, and

Shannon index were all greater for the milk of Japanese than those of Chinese and

Vietnamese cows. Moreover, relative abundances of Enterobacteriaceae and

Propionibacteriaceae were greater for Chinese and those of Streptococcaceae and

Methylobacteriaceae were greater for Vietnamese milk samples. Japanese milk

samples were characterized by low abundances of potential pathogens. It is yet unsure

if these differences can be due to the differences between AMS and parlor milking;

however, compared with regional difference examined in experiment 4, meteorological

difference could influence on the milk microbiota of dairy cows.

These experiments have revealed a number of factors affecting the milk

microbiota of dairy cows. The fact that substantial differences were seen in the milk

microbiota between manually collected and automatically collected samples should be

considered for both mastitis prevention and milk quality control. Although

Staphylococcaceae and Enterobacteriaceae were present at high proportions in udder

milk, the milk microbiota could be represented by Moraxellaceae and

Pseudomonadaceae in cooling tank, which may lead to an underestimate of mastitis

risk and irrelevant cow health management. Likewise, even though high abundances of

typical pathogens like Staphylococcaceae, Enterobacteriaceae, and Streptococcaceae

6

were shared in cows, the SCC was substantially different; hence, in addition to

monitoring of the milk microbiota, cow factors associated with infection and

inflammation need to be considered for better prevention and treatment. Our finding

that milk composition, especially milk urea nitrogen, was related with the abundance

of Moraxellaceae in the milk can suggest that appropriate protein-energy balance may

work as a measure to suppress pathogen contaminations of the milk.

7

CHAPTER 1 General Introduction

Bovine milk contains the nutrients needed for growth and development of the calf,

and is a resource of lipids, proteins, amino acids, vitamins and minerals. The lipids in

milk are emulsified in globules coated with membranes and proteins are in colloidal

dispersions as casein micelles. The casein micelles occur as colloidal complexes of

protein and salts, primarily calcium. Lactose and most minerals are in solution. Milk

composition has a dynamic nature, and the composition varies with stage of lactation,

age, breed, nutrition, and energy balance and health status of the udder. Milk contains

many different types of fatty acids. All these components make milk a nutrient rich

food item. Similarly, it has become a good environment for the growth of many kinds

of bacteria. Bacterial infection of the mammary gland results in an increase in the

concentration of somatic cells in cow’s milk.

1.1 Effects of Milk Contents on the Health of Dairy Cows

Many data show that milk content is related to mammary gland health and bacteria

in milk. Fat-to-protein ratio (F:P) may serve as a measure of cow’s energy balance

(Buttchereit et al., 2010). Patterns of changes in milk yield, F:P and SCC during

lactation was used to detect cases of mastitis in dairy cows (Jamrozik & Schaeffer,

2012). Heuer indicated that F:P was a risk factor for many diseases, including mastitis

(Heuer et al., 1999). Windig showed that cows with short and intensive peaks of SCS

related to infections with environmental pathogens exhibited an increase in F:P,

whereas infections caused by contagious pathogens resulted in decrease in F:P (Windig

et al., 2010). Kayano reported that Some mastitis pathogens, such as Streptococcus and

Escherichia coli, seem to be associated with long-term changes in fat, protein and MUN

(Kayano et al., 2018).

Milk triglyceride is synthesized from more than 400 different fatty acids, making

it the most complex of all-natural fats (Haug et al., 2007; Jensen, 2002). Almost all of

these acids exist in trace amounts, with only 1% or higher levels of about 15 acids.

Many factors are related to the changes of lipid content and fatty acid composition in

milk (Jensen, 2002; Palmquist et al., 1993). They may originate from animals, i.e.

genetics, lactation, mastitis and rumen fermentation, or they may be feed-related factors,

8

i.e. fiber and energy intake, dietary fat, seasonal and regional effects (Lindmark

Månsson, 2008). The fat content would be significant changed when the mastitis was

occurs (Kitchen, 1981).

Urea is an end product of protein metabolism. Rajala reported negative effects of

blood urea nitrogen or Mun on reproductive performance of dairy cows, and pointed

out that overfeeding CP resulted in reproductive stress (Rajala-Schultz et al., 2001).

However, others did not find that high concentrations of mirabilite had such a negative

effect on the reproductive capacity of dairy cows (Godden et al., 2001). Milk urea

nitrogen (MUN) is also used to monitor the utilization of protein in the gastrointestinal

tract of dairy cows in modern dairy farming. Rumen microorganisms break down feed

protein into ammonia. Rumen microorganisms can capture ammonia and use its

nitrogen to synthesize amino acids and proteins when their production is carried out in

the optimum amount of fermentable carbohydrates. However, in practice, especially in

high-yielding dairy cows, excessive ammonia occurs in the rumen, where ammonia is

absorbed through the rumen wall, penetrates into the blood, enters the liver, and the

liver cells convert it into urea in the urea cycle. Therefore, the urea content in milk

conveys detailed information about volume. Nitrogen, which is not used for the growth

or synthesis of milk proteins, is found in animal feed that is eaten every day. Feed

rationing contains too much protein, and excessive nitrogen is excreted into body fluids,

namely plasma (pun - plasma urea nitrogen), milk and urine (non-urine nitrogen).

Therefore, the level of urea in milk is a very useful indicator, which tells modern dairy

farmers that cows do not make full use of feed protein intake and excretion of excessive

nitrogen, or in the case of low protein content in feed. Actual use of information on

milk urea levels helps to assess the energy and protein balance of rations, reduce feed

costs and potential loss of nitrogen in dairy farms (Piotr Guliński et al., 2016; Wattiaux

1994). MUN concentration could be affected by season. Guli showed high levels of

protein associated with milk production season and feed. In their study, milk produced

in summer had the highest urea content (225 mg/l), while milk produced in winter had

the lowest urea content (172 mg/l) (P Guliński et al., 2008).

1.2 Progress and shortcomings of AMS milking machine

Since 1992, Automatic milking systems (AMS) were started to use in the

Netherlands and become more and more popular from Europe to all over the world

9

(Jacobs & Siegford, 2012; Shortall et al., 2016). The milk collection separates from

human participation, it has brought a vast reduce of stress and fear response (Innocente

& Biasutti, 2013). Moreover, the use of AMS enables cows to be milked whenever they

want; hence, compared with regular twice per day milking, milk production may

increase as the milking frequency of individual cows increases (Kruip et al., 2002). On

the same time, mastitis monitoring system is well going for quality assurance of milk

(Castro et al., 2015; Jacobs & Siegford, 2012; Kamphuis et al., 2010; Mollenhorst et

al., 2012). The AMS is equipped with instruments to detect individual affected breast

seasons. Milk conductance and color analysis were used to detect affected milk

(Hovinen et al., 2006). These systems make it possible to detect affected breast seasons

and separate milk from those affected quarters. However, udder health is also a

worrying problem on farms (Dohmen et al., 2010; Hovinen & Pyörälä, 2011). Honvinen

reported somatic cell count (SCC) was increased after AMS introduction first year and

continued to be higher (Hovinen et al., 2009). And, under AMS management, Mastitis

morbidity did not change in Primiparous cows, but Multiparous cows were increased.

it cannot be ruled out that the hypothetical healthy composite milk may contain high

SCC milk from various breast regions, which may affect bulk canned milk during

storage due to enzyme activity (Berglund et al., 2004). And the alert may increase the

antibiotic using for mastitis treatment (Meijering et al., 2004).

1.3 Bacteria in Milk

Microbiome of milk has always been attention. From prevent and treatment of

mastitis to prevent spoilage and quality identification and then to process of dairy

products (e.g. cheese, yogurt), every step was correlated with microflora in milk. A

wide variety study of bacterial of milk was contributed with culture-dependent and -

independent to investigate the source of milk bacteria (Dogan et al., 2006; Jonghe et al.,

2011; Thomas et al., 2003). In the vast majority of studies, there are two main categories:

cow health (mastitis) and milk quality (spoilage).

1.3.1 Mastitis

Mastitis is an inflammation of the breast, which is a common and expensive

disease in many dairy cattle around the world (Halasa et al., 2007). Milk cows

experienced milk loss and poor milk quality (Ogola et al., 2007). Milk loss is a major

10

component of the cost of mastitis. Poor milk quality, such as high somatic cell count

(SCC) and changes in milk ingredients, may damage profits because it may lower milk

prices. It is very important for dairy industry, especially for cheese production and

quality (Blum et al., 2014). Clinical mastitis may be associated with changes in fat,

protein and MUM in milk (Kayano et al., 2018). The changes in fat and protein may be

caused by clinical mastitis. Although some of them are not pathogen specific, they show

high fat and low protein content (Coulon et al., 2002; Jamrozik & Schaeffer, 2012); on

the contrary, low fat content is caused by streptococcal infection (Bezman et al., 2015).

Other studies have shown no significant changes in fat and protein content due to

mastitis (Botaro et al., 2014; Tomazi et al., 2015).

Mastitis affects total milk production and changes milk composition and technical

availability. In dairy cows, somatic cell count (SCC) is a useful predictor of subclinical

mastitis. It is an important component of milk in terms of quality, hygiene and mastitis

control (Harmon 1994). The increase of SCC in milk was also related to the change of

protein quality, fatty acid composition, lactose, ion and mineral concentration, the

increase of enzyme activity and the higher pH value of raw milk (Auldist et al., 1996;

Coulon et al., 2002). SCC is lowest during the winter, highest during the summer, and

greater for multiparous than primiparous cows (Olde Riekerink et al., 2007). Both

contagious and environmental pathogens contribute to the high SCC observed in

summer.

A variety of bacterial pathogens are known to be related to mastitis,

Staphylococcus aureus and Streptococcus agalactiae are regarded as the most common

contagious pathogens, and coagulase-negative Staphylococci, Escherichia coli,

Streptococcus uberis, and Streptococcus dysagalactiae are regarded as the most

common environmental pathogens (Bradley, 2002). Kayano reported that Some

mastitis pathogens, such as Streptococcus and Escherichia coli, seem to be associated

with long-term changes in fat, protein and MUN (Kayano et al., 2018). And long-term

milk losses were caused by the contamination of Escherichia coli and Streptococcus

spp. (Bezman et al., 2015; Kayano et al., 2018). However, approximately 10–40% of

clinical mastitis cases are culture negative in routine clinical assays (Makovec & Ruegg,

2003). Chryseobacterium spp. are opportunistic bacteria in raw milk;(Hagi et al., 2013)

found these bacteria in the milk of healthy cows, and (Kuang et al., 2009) detected these

bacteria in the milk of cows with mastitis. Based on NGS, (Zhang et al., 2014) detected

Chryseobacterium spp. as the most abundant species, whereas (Oikonomou et al., 2014)

11

detected these bacteria as rather minor species in raw milk. Enterococcus spp. are also

often found in raw milk samples, and the source of contamination is thought to be

milking equipment rather than the animal body and faeces (Kagkli et al., 2007).

Propionibacterium acnes is a regular inhabitant of animal skin, and (Oikonomou et al.,

2014) detected this species by NGS as the predominant species in healthy milk samples.

Despite the finding that Burkholderia spp. were more frequent in clinically diseased

milk than in healthy milk (Kuehn et al., 2013).

1.3.2 Spoilage

Certainly, pasteurized milk is as safe as a house, however, the residual secretion

of bacterial in raw milk still induce to spoilage and quality defects (Kable et al., 2016).

As a well-known measure, refrigerated storage of raw milk is widely used for prolong

storage time and mesophilic bacteria spoilage prevention (Marchand, et al., 2017).

Indeed, the low temp storage due to an outgrowth of psychrophilic bacteria in raw milk.

The predominant psychrophilic bacterial secrete heat-stable enzymes while milk was

heat-treated (e.g. UHT). Lafarge report that psychrophilic bacterial were significant

increase in low temp storage (Decimo et al., 2014; Machado et al., 2015; Marchand et

al., 2009; Ogier et al., 2004). Pseudomonas spp. are well-known psychrotrophic

bacteria and are found more often during the winter in raw milk samples. (Hagi et al.,

2013; Uraz & Citak, 1998) Pseudomonadaceae family and Moraxellaceae family

would produce heat-stable extracellular enzymes aprX-lipA, principally pseudomonas

secreted proteases and Acinetobacter, microbacterium, pseudomonas and Enterobacter

secreted lipases (Vithanage et al., 2016). The proteases casein micelles to casein gather

and precipitate to disequilibria milk gelatinize and lipase can hydrolyze acylglycerol to

release fatty acids like unpleasant flavors short-chain fatty acids (e.g. butyric, caprylic

and caproic acids) and lead to soapy taste medium-chain fatty acids (Chen et al., 2003;

Machado et al., 2017). Therefore, the increase of somatic cell count may reduce the

storage time of milk. Therefore, (Jonghe et al., 2011) suggested that proteolysis and

lipolysis was an important spoilage potential as determined. N2 gas flushing was used

to inhibit psychrotrophic pseudomonas increasing at low temp. However, the

psychrotrophic bacterial source is not clear yet (Gschwendtner et al., 2016). On the

other hand, A study using next-generation sequencing (NGS) indicated that the

proportion of Pseudomonas spp. in raw milk was greater in healthy cows than in cows

12

exhibiting clinical disease (Kuehn et al., 2013). Similarly, Barbano ma’s paper was

found that the speed of spoilage of milk with high somatic cell count was much higher

than that of milk with low somatic cell counts (Barbano et al., 2006). Ahmed Gargouri’s

paper found that the polymorphonuclear leukocytes (PMN) were significantly

correlated (P<0.001) with total SCC, and lipolysis level in milk. The PMN was

suggested that it can response to sufficient lipolysis of milk fat (Gargouri et al., 2008).

In any case, the high SCC would accelerate the flavor lost or spoilage in milk.

REFERENCE

Auldist, M. J., Coats, S., Sutherland, B. J., Mayes, J. J., McDowell, G. H., & Rogers,

G. L. (1996). Effects of somatic cell count and stage of lactation on raw milk

composition and the yield and quality of Cheddar cheese. Journal of Dairy

Research, 63(2), 269–280. https://doi.org/10.1017/S0022029900031769

Barbano, D. M., Ma, Y., & Santos, M. V. (2006). Influence of raw milk quality on

fluid milk shelf life. Journal of Dairy Science, 89, E15–E19.

https://doi.org/10.3168/jds.S0022-0302(06)72360-8

Berglund, I., Pettersson, G., Ostensson, K., & Svennersten-Sjaunja, K. (2004).

Frequency of individual udder quarters with elevated CMT scores in cows’ milk

samples with low somatic cell counts. The Veterinary Record, 155(7), 213.

https://doi.org/10.1136/VR.155.7.213

Bezman, D., Lemberskiy-Kuzin, L., Katz, G., Merin, U., & Leitner, G. (2015).

Influence of intramammary infection of a single gland in dairy cows on the

cow’s milk quality. Journal of Dairy Research, 82(03), 304–311.

https://doi.org/10.1017/S002202991500031X

Blum, S. E., Heller, E. D., & Leitner, G. (2014). Long term effects of Escherichia coli

mastitis. The Veterinary Journal, 201(1), 72–77.

https://doi.org/10.1016/J.TVJL.2014.04.008

Botaro, B. G., Cortinhas, C. S., Dibbern, A. G., Silva, L. F. P. e., Benites, N. R., &

dos Santos, M. V. (2014). Staphylococcus aureus intramammary infection affects

milk yield and SCC of dairy cows. Tropical Animal Health and Production,

47(1), 61–66. https://doi.org/10.1007/s11250-014-0683-5

Bradley, A. J. (2002). Bovine mastitis: An evolving disease. The Veterinary Journal,

164(2), 116–128. https://doi.org/10.1053/TVJL.2002.0724

13

Buttchereit, N., Stamer, E., Junge, W., & Thaller, G. (2010). Evaluation of five

lactation curve models fitted for fat: protein ratio of milk and daily energy

balance. Journal of Dairy Science, 93(4), 1702–1712.

https://doi.org/10.3168/jds.2009-2198

Castro, A., Pereira, J. M., Amiama, C., & Bueno, J. (2015). Mastitis diagnosis in ten

Galician dairy herds (NW Spain) with automatic milking systems. Spanish

Journal of Agricultural Research, 13(4). https://doi.org/10.5424/sjar/2015134-

7482

Chen, L., Daniel, R. M., & Coolbear, T. (2003). Detection and impact of protease and

lipase activities in milk and milk powders. International Dairy Journal, 13(4),

255–275. https://doi.org/10.1016/S0958-6946(02)00171-1

Coulon, J.-B., Gasqui, P., Barnouin, J., Ollier, A., Pradel, P., & Pomiès, D. (2002).

Effect of mastitis and related-germ on milk yield and composition during

naturally-occurring udder infections in dairy cows. Anim. Res, 51.

https://doi.org/10.1051/animres:2002031

Decimo, M., Morandi, S., Silvetti, T., & Brasca, M. (2014). Characterization of

Gram-negative psychrotrophic bacteria isolated from Italian bulk tank milk.

Journal of Food Science, 79(10), M2081–M2090. https://doi.org/10.1111/1750-

3841.12645

Dogan, B., Klaessig, S., Rishniw, M., Almeida, R. A., Oliver, S. P., Simpson, K., &

Schukken, Y. H. (2006). Adherent and invasive Escherichia coli are associated

with persistent bovine mastitis. Veterinary Microbiology, 116(4), 270–282.

https://doi.org/10.1016/j.vetmic.2006.04.023

Dohmen, W., Neijenhuis, F., & Hogeveen, H. (2010). Relationship between udder

health and hygiene on farms with an automatic milking system. Journal of Dairy

Science, 93(9), 4019–4033. https://doi.org/10.3168/jds.2009-3028

Gargouri, A., Hamed, H., & ElFeki, A. (2008). Total and differential bulk cow milk

somatic cell counts and their relation with lipolysis. Livestock Science, 113(2–

3), 274–279. https://doi.org/10.1016/j.livsci.2007.05.007

Godden, S. M., Kelton, D. F., Lissemore, K. D., Walton, J. S., Leslie, K. E., &

Lumsden, J. H. (2001). Milk urea testing as a tool to monitor reproductive

performance in ontario dairy herds. Journal of Dairy Science, 84(6), 1397–1406.


Gschwendtner, S., Alatossava, T., Kublik, S., Fuka, M. M., Schloter, M., & Munsch-

14

Alatossava, P. (2016). N2 gas flushing alleviates the loss of bacterial diversity

and inhibits psychrotrophic pseudomonas during the cold storage of bovine raw

milk. PLoS ONE, 11(1), 1–17. https://doi.org/10.1371/journal.pone.0146015

Guliński, P., Młynek, K., & Weterynaryjna, E. S.M. (2008). Milk urea concentration

changes depending on selected environmental factors. Cabdirect.Org. Retrieved

from https://www.cabdirect.org/cabdirect/abstract/20083095410

Guliński, P., Salamończyk, E., & Młynek, K. (2016). Improving nitrogen use

efficiency of dairy cows in relation to urea in milk - A review. Animal Science

Papers and Reports, 34(1), 5–24.

Hagi, T., Sasaki, K., & Aso, H. (2013). Adhesive properties of predominant bacteria

in raw cow’ s milk to bovine mammary gland epithelial cells. Folia

Microbiologica, 58(6),515–522. https://doi.org/10.1007/s12223-013-0240-z

Halasa, T., Huijps, K., Østerås, O., & Hogeveen, H. (2007). Economic effects of

bovine mastitis and mastitis management: A review. Veterinary Quarterly, 29(1),

18–31. https://doi.org/10.1080/01652176.2007.9695224

Harmon, R. J. (1994). Physiology of mastitis and factors effecting somatic cell counts.

Journal of Dairy Science, 77(7), 2103–2112. https://doi.org/10.3168/JDS.S0022-

0302(94)77153-8

Haug, A., Hostmark, A., & Harstad, O. (2007). Bovine milk in human nutrition - a

review. Lipids in Health and Disease, 6(1), 25. https://doi.org/10.1186/1476-

511X-6-25

Heuer, C., Schukken, Y. H., & Dobbelaar, P. (1999). Postpartum body condition score

and results from the First test day milk as predictors of disease, fertility, yield,

and culling in commercial dairy herds. Journal of Dairy Science, 82(2), 295–304.


Hovinen, M., Aisla, A. M., & Pyörälä, S. (2006). Accuracy and reliability of mastitis

detection with electrical conductivity and milk colour measurement in automatic

milking. Acta Agriculturae Scandinavica A: Animal Sciences, 56(3–4), 121–127.

https://doi.org/10.1080/09064700701216888

Hovinen, M., & Pyörälä, S. (2011). Invited review: Udder health of dairy cows in

automatic milking. Journal of Dairy Science, 94(2), 547–562.

https://doi.org/10.3168/jds.2010-3556

Hovinen, M., Rasmussen, M. D., & Pyörälä, S. (2009). Udder health of cows

changing from tie stalls or free stalls with conventional milking to free stalls with

15

either conventional or automatic milking. Journal of Dairy Science, 92(8), 3696–

3703. https://doi.org/10.3168/jds.2008-1962

Innocente, N., & Biasutti, M. (2013). Automatic milking systems in the protected

designation of origin montasio cheese production chain: Effects on milk and

cheese quality. Journal of Dairy Science, 96(2), 740–751.

https://doi.org/10.3168/jds.2012-5512

Jacobs, J. A., & Siegford, J. M. (2012). Lactating dairy cows adapt quickly to being

milked by an automatic milking system. Journal of Dairy Science, 95(3), 1575–

1584. https://doi.org/10.3168/jds.2011-4710

Jamrozik, J., & Schaeffer, L. R. (2012). Test-day somatic cell score, fat-to-protein

ratio and milk yield as indicator traits for sub-clinical mastitis in dairy cattle.

Journal of Animal Breeding and Genetics, 129(1), 11–19.

https://doi.org/10.1111/j.1439-0388.2011.00929.x

Jensen, R. G. (2002). The composition of bovine milk lipids: January 1995 to

December 2000. Journal of Dairy Science, 85(2), 295–350.

https://doi.org/10.3168/JDS.S0022-0302(02)74079-4

Jonghe, V. De, Coorevits, A., Hoorde, K. Van, Messens, W., Landschoot, A. Van,

Vos, P. De, & Heyndrickx, M. (2011). Influence of storage conditions on the

growth of Pseudomonas species in refrigerated raw milk. Appl Environ

Microbiol, 77(2):460-470. https://doi.org/10.1128/AEM.00521-10

Kable, M. E., Srisengfa, Y., Laird, M., Zaragoza, J., McLeod, J., Heidenreich, J., &

Marco, M. L. (2016). The core and seasonal microbiota of raw bovine milk in

tanker trucks and the impact of transfer to a milk processing facility. MBio, 7(4),

1–13. https://doi.org/10.1128/mBio.00836-16

Kagkli, D. M., Vancanneyt, M., Hill, C., Vandamme, P., & Cogan, T. M. (2007).

Enterococcus and Lactobacillus contamination of raw milk in a farm dairy

environment. International Journal of Food Microbiology, 114(2), 243–251.

https://doi.org/10.1016/J.IJFOODMICRO.2006.09.016

Kamphuis, C., Mollenhorst, H., Heesterbeek, J. A. P., & Hogeveen, H. (2010).

Detection of clinical mastitis with sensor data from automatic milking systems is

improved by using decision-tree induction. Journal of Dairy Science, 93(8),

3616–3627. https://doi.org/10.3168/jds.2010-3228

Kayano, M., Itoh, M., Kusaba, N., Hayashiguchi, O., Kida, K., Tanaka, Y., … Gröhn,

Y. T. (2018). Associations of the first occurrence of pathogen-specific clinical

16

mastitis with milk yield and milk composition in dairy cows. Journal of Dairy


Kitchen, B. J. (1981). Review of the progress of dairy science: Bovine mastitis: Milk

compositional changes and related diagnostic tests. Journal of Dairy Research,

48(1), 167–188. https://doi.org/10.1017/S0022029900021580

Kruip, T. a M., Morice, H., Robert, M., & Ouweltjes, W. (2002). Robotic milking and

its effect on fertility and cell counts. Journal of Dairy Science, 85(10), 2576–

2581. https://doi.org/10.3168/jds.S0022-0302(02)74341-5

Kuang, Y., Tani, K., Synnott, A. J., Ohshima, K., Higuchi, H., Nagahata, H., & Tanji,

Y. (2009). Characterization of bacterial population of raw milk from bovine

mastitis by culture-independent PCR-DGGE method. Biochemical Engineering

Journal, 45(1), 76–81. https://doi.org/10.1016/j.bej.2009.02.010

Kuehn, J. S., Gorden, P. J., Munro, D., Rong, R., Dong, Q., Plummer, P. J., …

Phillips, G. J. (2013). Bacterial community profiling of milk samples as a means

to understand culture-negative bovine clinical mastitis, 8(4).

https://doi.org/10.1371/journal.pone.0061959

Lindmark Månsson, H. (2008). Fatty acids in bovine milk fat. Food and Nutrition

Research, 52, 4–7. https://doi.org/10.3402/fnr.v52i0.1821

Machado, S. G., Baglinière, F., Marchand, S., Van Coillie, E., Vanetti, M. C. D., De

Block, J., & Heyndrickx, M. (2017). The biodiversity of the microbiota

producing heat-resistant enzymes responsible for spoilage in processed bovine

milk and dairy products. Frontiers in Microbiology, 8, 302.

https://doi.org/10.3389/FMICB.2017.00302

Machado, S. G., da Silva, F. L., Bazzolli, D. M. S., Heyndrickx, M., Costa, P. M. de

A., & Vanetti, M. C. D. (2015). Pseudomonas spp. and Serratia liquefaciens as

predominant spoilers in cold raw milk. Journal of Food Science, 80(8), M1842–

M1849. https://doi.org/10.1111/1750-3841.12957

Makovec, J. A., & Ruegg, P. L. (2003). Results of milk samples submitted for

microbiological examination in wisconsin from 1994 to 2001. Journal of Dairy

Science, 86(11), 3466–3472. https://doi.org/10.3168/jds.S0022-0302(03)73951-4

Marchand, S., Duquenne, B., Heyndrickx, M., Coudijzer, K., & De Block, J. (2017).

Destabilization and off-flavors generated by Pseudomonas proteases during or

after UHT-processing of milk. International Journal of Food Contamination,

4(1), 2. https://doi.org/10.1186/s40550-016-0047-1

17

Marchand, S., Heylen, K., Messens, W., Coudijzer, K., De Vos, P., Dewettinck,

K., … Heyndrickx, M. (2009). Seasonal influence on heat-resistant proteolytic

capacity of Pseudomonas lundensis and Pseudomonas fragi, predominant milk

spoilers isolated from Belgian raw milk samples. Environmental Microbiology,

11(2), 467–482. https://doi.org/10.1111/j.1462-2920.2008.01785.x

Meijering, A., Hogeveen, H., & de Koning, C. J. A. M. (Eds.). (2004). Automatic

milking, a better understanding. The Netherlands: Wageningen Academic

Publishers, 544. https://doi.org/10.3920/978-90-8686-525-3

Mollenhorst, H., Rijkaart, L. J., & Hogeveen, H. (2012). Mastitis alert preferences of

farmers milking with automatic milking systems. Journal of Dairy Science,

95(5), 2523–2530. https://doi.org/10.3168/jds.2011-4993

Ogier, J., Girard, V., Leveau, J., & Gruss, A. (2004). Raw cow milk bacterial

population shifts attributable to refrigeration. Appl Environ Microbiol,

70(9):5644-5650. https://doi.org/10.1128/AEM.70.9.5644

Ogola, H., Shitandi, A., & Nanua, J. (2007). Effect of mastitis on raw milk

compositional quality. J Vet Sci, 8(3):237-242.

Oikonomou, G., Bicalho, M. L., Meira, E., Rossi, R. E., Foditsch, C., Machado, V.

S., … Bicalho, R. C. (2014). Microbiota of cow’s milk; distinguishing healthy,

sub-clinically and clinically diseased quarters. PLoS ONE, 9(1).


Olde Riekerink, R. G. M., Barkema, H. W., & Stryhn, H. (2007). The effect of season

on somatic cell count and the incidence of clinical mastitis. Journal of Dairy


Palmquist, D. L., Denise Beaulieu, A., & Barbano, D. M. (1993). Feed and animal

factors influencing milk fat composition. Journal of Dairy Science, 76(6), 1753–


Rajala-Schultz, P. J., Saville, W. J. A., Frazer, G. S., & Wittum, T. E. (2001).

Association between milk urea nitrogen and fertility in Ohio Dairy Cows.

Journal of Dairy Science, 84(2), 482–489. https://doi.org/10.3168/jds.S0022-

0302(01)74498-0

Shortall, J., Shalloo, L., Foley, C., Sleator, R. D., & O’Brien, B. (2016). Investment

appraisal of automatic milking and conventional milking technologies in a

pasture-based dairy system. Journal of Dairy Science, 99(9), 7700–7713.

https://doi.org/10.3168/jds.2016-11256

18

Thomas, A., Sachse, K., Farnir, F., Dizier, I., Mainil, J., & Linden, A. (2003).

Adherence of Mycoplasma bovis to bovine bronchial epithelial cells. Microbial

Pathogenesis, 34(3), 141–148. https://doi.org/10.1016/S0882-4010(03)00003-2

Tomazi, T., Gonçalves, J. L., Barreiro, J. R., Arcari, M. A., & Santos, M. V. (2015).

Bovine subclinical intramammary infection caused by coagulase-negative

staphylococci increases somatic cell count but has no effect on milk yield or

composition. Journal of Dairy Science, 98(5), 3071–3078.

https://doi.org/10.3168/jds.2014-8466

Uraz, G., & Citak, S. (1998). The isolation of Pseudomonas and other Gram (-)

psychrotrophic bacteria in raw milks. Journal of Basic Microbiology, 38(2), 129–

134. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/9637013

Vithanage, N. R., Dissanayake, M., Bolge, G., Palombo, E. A., Yeager, T. R., &

Datta, N. (2016). Biodiversity of culturable psychrotrophic microbiota in raw

milk attributable to refrigeration conditions, seasonality and their spoilage

potential. International Dairy Journal, 57, 80–90.

https://doi.org/10.1016/j.idairyj.2016.02.042

Wattiaux, M. A., Combs, D. K., & Shaver, R. D. (1994). Lactational responses to

ruminally undegradable protein by dairy cows fed diets based on alfalfa silage. J

Dairy Sci, 77(6):1604-1617.

Windig, J. J., Ouweltjes, W., ten Napel, J., de Jong, G., Veerkamp, R. F., & De Haas,

Y. (2010). Combining somatic cell count traits for optimal selection against

mastitis. Journal of Dairy Science, 93(4), 1690–1701.

https://doi.org/10.3168/jds.2009-2052

Zhang, R., Huo, W., Zhu, W., & Mao, S. (2014). Characterization of bacterial

community of raw milk from dairy cows during subacute ruminal acidosis

challenge by high-throughput sequencing. J Sci Food Agric, 95(5):1072-9.

https://doi.org/10.1002/jsfa.6800

19

CHAPTER 2 Monitoring the microbiota and Nutrient

composition in dairy cow milk collected from a herd

managed by automatic milking systems

2.1 INTRODUCTION

In recent years, the dairy industry has experienced many technological

advancements. Among these advancements, implementation of Automatic milking

systems (AMS) is one of the most important events occurring within the last 20 years.

AMS have greatly affected milk production, milk quality, cow behaviour, cow welfare,

herd management, and herd health (Jacobs & Siegford, 2012). In the milk production

process, all operations follow the system settings completely automated by the machine,

greatly reducing the risk of bacterial contamination caused by manual operation. At the

same time, the AMS will automatically collect individual dairy cow samples, not only

for the convenience of individual management of cattle health, but also for the quality

of milk and dairy products to provide effective sample support. However, the majority

of published research still has examined the microbiota of milk in tanks (Frétin et al.,

2017; Kable et al., 2016; Kim et al., 2017; Li et al., 2018; Meng et al., 2017; Quigley

et al., 2013; Rodrigues et al., 2017; Zhang et al., 2014). Data on individual milk quality

and milk flora are still few.

The bacteria in milk play a direct role in uber health as well as milk and dairy

products. Some bacterial pathogens are known to be related to mastitis. Staphylococcus

aureus and Streptococcus agalactiae are regarded as the most common contagious

pathogens, and coagulase-negative Staphylococci, Escherichia coli, Streptococcus

uberis, and Streptococcus dysagalactiae are regarded as the most common

environmental pathogens (Bradley, 2002). Some bacteria in milk, including

Lactococcus, Lactobacillus, Streptococcus, Propionibacterium as well as fungi can

directly affect the flavor, and organoleptic properties of dairy products (Wouters et al.,

2002). Psychrophilic bacterial, including Pseudomonas spp. and Acinetobacter spp.,

can grow in milk at low temperature. These bacteria can produce proteases and lipases

even after pasteurization, leading to nutrition and organoleptic changes associated with

spoilage (Ribeiro Júnior et al., 2017; Vithanage et al., 2016). The quality of milk,

20

including somatic cell count (SCC) and total bacterial count (TBC), will obviously

affect the milk shelf life and the yield of cheese. (Auldist et al., 1996; Barbano et al.,

2006). Moreover, Gabriel Leitner suggests that the quality of milk that affects cheese

production may be due to subclinical uber infections when milk is mixed in milk tank

(Leitner et al., 2008). Therefore, it is necessary to investigate the flora of individual

milk with different milk quality levels.

Milk microbiota has been shown to be influenced by many factors including region,

season, cowshed environment, and the hygienic management of milking (Doyle et al.,

2017b; Kable et al., 2016; Kim et al., 2017; Latorre et al., 2010) It is worth noting that

the content in milk can also affect the quality of dairy products. Milk fat and protein

content will directly affect the yield of cheese (Verdier-Metz et al., 2001). And, the

amount of urea in milk affects the taste of cheese (Martin et al., 1997). Interestingly,

even in the same feeding conditions, the components in different milk is not the same.

For example, milk protein and fat in different seasons, parity and Lactation period

content will be significantly different (Yoon et al., 2004). In addition, MUN as an index

of dietary crude protein intake (Nousiainen et al., 2004), also different levels in different

seasons (Hwang et al., 2000; Yoon et al., 2004). However, there was less paper show

the effects of content on the bacterial flora in milk.

In this study, we examined the microbiota and milk contents of dairy cow milk

collected from a herd managed by AMS. Because both microbiota and milk contents

can be influenced by season, we performed a 1-year survey using test-day milk samples.

Select and carry out microbial analysis of milk with different levels of quality (SCC)

per month. Microbiota was analysed by MiSeq, and the total bacterial population was

measured by qPCR. The objectives of this study were to provide information about the

milk microbiota of the cows managed by AMS and to clarify changes in the milk

microbiota in relation to milk contents and varying levels of milk quality.

2.2 MATERIALS AND METHODS

2.2.1 Milk samples

Milk samples were collected from a Holstein herd managed by AMS (Lely

Astronout A4, Cornes AG. Ltd., Eniwa, Japan) at Okayama Prefecture Livestock

Research Institute. The cows were fed total mixed ration silage throughout the year and

their milk samples were collected once a month for the herd test. Concentrations of dry

21

matter (DM), crude protein (N × 6.25), and total digestible nutrients in the total mixed

ration silage were 500–600 g kg–1, 160–180 g kg–1DM, and 720–740 g kg–1DM,

respectively, and dairy cows showed a high level of milk production without other

supplements. (Han et al., 2014) The test-day samples mechanically taken by the AMS

were collected every 2 months from February 2014 to December 2014. Milk samples

were transferred to a 1.5-mL sterile centrifuge tube and stored on ice until transported

to the university. The tubes were stored at –30°C until analysis. SCC and milk

component were determined using a CombiFoss FT+ (Foss Alle, Hillerød, Denmark)

after milk treated with Bronopol (Chemical Land○R). A total of 201 samples were

collected in this 1-year survey.

2.2.2 Preparation of bacterial DNA

Totally 36 samples of somatic cells with varying degrees in each sampling time,

including less than 100,000 cells/ml, 101,000 to 300,000 cells/ml, and greater than

301,000 cells/ml were selected for analysis. Bacterial pellets were obtained by

centrifugation at 8000 × g for 15 min, and then washed once with 1 mL of solution

containing 0.05 M D-glucose, 0.025 M Tris-HCl (pH 8.0), and 0.01 M sodium EDTA

(pH 8.0). After centrifugation at 15,000 × g for 2 min, bacterial cells were lysed with

180 μL of lysozyme solution (20 g/L lysozyme, 0.02 M Tris-HCl [pH 8.0], 0.002 M

sodium EDTA [pH 8.0], 1.2 g/L Triton X-100) at 37°C for 1 h. Subsequent bacterial

DNA purification was performed using a commercial kit (DNeasy Blood & Tissue Kit;

Qiagen, Germantown, MD, USA) according to the manufacturer’s

recommendations(Ni et al., 2017).

2.2.3 qPCR

qPCR was carried out on a MiniOpticon System (Bio-Rad Laboratories, Inc.,

Tokyo, Japan). For quantification of both total bacteria, 2 μL of DNA solution was

added to 23 μL of a PCR mixture containing 12.5 μL of KAPA SYBR FAST Master

Mix (Kapa Biosystems, Inc., Wilmington, MA, USA) and 1.0 μL of each primer (8 μM,

357f and 517r, as described above). A dilution series of plasmids carrying the nearly

full-length 16S rDNA gene of E. coli was used as known concentrations of the standard

plasmid. The cycle parameters for the total bacterial assay were as follows: 30 s at 95°C

and 35 cycles of 15 s at 95°C, 20 s at 60°C, and 30 s at 72°C. The copy number of the

22

standard plasmid was calculated using the molecular weight of the nucleic acid and the

length (base pairs) of the cloned plasmid.

2.2.4 Illumina MiSeq sequencing

The PCR amplification using primers targeting the V4 region of the 16S rRNA

genes (forward: 5ʹ-ACACTCTTTCCCTACACGACGCTCTTCCGATCTGTGCCAG

CMGCCGCGGTAA-3′; reverse: 5′-GTGACTGGAGTTCAGACGTGTGCTCTT

CCGATCTGGACTACHVGGG TWTCTAAT-3′) was employed (Tang et al. 2017).

The PCR protocol was as follows: initiation at 94°C for 2 min and followed by 25 cycles

of 94°C for 30 s, 50°C for 30 s, 72°C for 30 s, and a final elongation of 72°C for 5 min.

The products were purified using Fast Gene Gel/PCR Extraction Kit (NIPPON

Genetics Co., Ltd., Tokyo, Japan) and moved to a second round of PCR with adapter-

attached primers. The second PCR protocol was as follows: initiation at 94°C for 2 min

followed by 10 cycles of 94°C for 30 s, 59°C for 30 s, 72°C for 30 s and a final

elongation of 72°C for 5 min. The PCR products were again purified as described above.

The purified DNA was then ligated to the 16S rRNA amplicons prior to 2x250-bp

paired-end sequencing performed on an Illumina MiSeq platform at FASMAC Co., Ltd.

(Kanagawa, Japan).

Raw sequences were processed using QIIME (version 1.9.2) running the virtual

box microbial ecology pipeline. Each pair raw sequences data was joined longer than

200bp after cut off the low-quality sequences (Q<30). Chimeric sequences were

identified with USEARCH and removed. The remaining DNA sequences were grouped

into OTU with 97% matched with the closed-reference OTU picking method in QIIME,

with default settings. Both chimera checking and OTU picking used Greengenes 13.8

as the reference database. Sequences were aligned using PyNAST.

2.2.5 Statistical analysis

Quantitative NGS data were subjected to one-way analysis of variance. Tukey’s

multiple range test was performed using the JMP software (version 11; SAS Institute,

Tokyo, Japan) to examine the differences between sampling month and SCC degrees.

Data for microbiota were used to perform principle coordinate analysis and heatmap

construction using Primer version 7 with Permanova+ add-on software (Primer-E,

Plymouth Marine Laboratory, Plymouth, UK). Alpha diversity indices, i.e. Evenness,

23

observed otus, and Shannon indices, were rarefied to an equal depth of 30,000

sequences by QIIME. LEfSe Analysis of Seasonal Change by Galaxy

(http://huttenhower.sph.harvard.edu/galaxy). The correlation between the milk

components and relative genera were calculated by Spearman’s rank correlation

coefficient by SPSS (Statistics 24).

Figure 2.1 Seasonal variation of milk composition by one-way ANOVA. Feb

n=36; Apr n=52; Jun n=22; Aug n=31; Oct n=37; Dec n=23

2.3 RESULTS

A total of 201 samples were collected and analyzed in different months in cattle

farms that supplied the same TMR feed throughout the year. The protein content was

the highest in December while the MUN value was the lowest in the whole year. In

spring and autumn (April and October), the content of SNF in milk was the highest

(fig1). Among 201 milk samples, SCC of thirty-seven samples (18.4%) more than

300,000 cells mL-1, moreover, SCC of nine samples more than 1,000,000 cells mL-1.

24

However, no matter milk yield, milk fat content, or SCC in different months did not

change significantly (P > 0.05). In addition, there was a significant correlation between

milk components in person analysis. SCC was positively correlated with milk yield,

milk fat and milk protein (P < 0.05). At the same time, SCC in milk increased with the

increase of month age and parity of dairy cows. There was a significant positive

correlation between month age and Parity of dairy cows and milk production (fig2 a, P

< 0.01). There was a linear positive correlation between total bacterial count (TBC) and

SCC in milk collected by Automatic milking machine.

Figure 2.2 (A) Relation between Milk Indicators (n=201). (B) The somatic cell

count (SCC) in milk has a significant linear positive relationship with the total

bacterial count (TBC) (n=60). Data was calculated by Spearman’s rank correlation

coefficient and plotted in RStudio (Corrplot packages). SNF: Solid Non-Fat; MUN:

Milk Urea Nitrogen; MA: Month Age; DIM: Day in Milk’ *: p<0.05; **:P<0.01;

***:P<0.001

In MiSeq analysis, the average non-chimeric sequence reads in all samples was

38746. The highest bacterial species richness index (Chao1) was found in December

25

milk samples. Whether it is Shannon (Microbial diversity index) or Simpson (Flora

evenness), the milk collected in February is the highest and the lowest in August among

the whole year milk samples (fig 3). Firmicutes, Proteobacteria, Bacteroidetes,

Actinobacteria and Fusobacteria are the major Phylum in milk. From February to

August, the temperature of the environment gradually increased, and the content of

Firmicutes in milk samples increased significantly, and reached the highest level in

August. Meanwhile, the content of Proteobacteria decreased. Furthermore, the

Bacteroidetes proportion in milk samples were less than 1% in June, August and

October, more than 10 times lower than other months (fig 4a). The proportion of

bacteria in milk changed significantly in different months. In LEfSe analysis, the

contents of 11 species of family，such as Pseudomonadaceae and Enterobacteriaceae,

were significantly higher than those in other months. Streptococcaceae and

Microbacteriaceae are abundant in April milk samples. Enterococcaceae,

Planococcaceae and Bacillaceae were most in June milk than others month. The

contents of Carnobacteriaceae in August and Moraxellaceae in October were

significantly higher than those in other months, respectively. In December , The

contents of nine fungi including Staphylococcaceae, Micrococcaceae ， and

Ruminococcaceae were significantly higher than those in other months (fig 4b). There

were significant differences between warm seasonal milks (Jun, Aug and Oct) and cool

seasonal milks (Feb, Apr and Dec) in bacterial flora (fig 4c).

Figure 2.3 alpha diversity index of milk microbiota by one-way ANOVA.

26

Nutritional components, quality indicators and milk production factors in milk are

related to the proportion of bacteria in milk. There was a significant positive correlation

between TBC and protein content in milk (correlation coefficient 0.45; P < 0.01). When

TBC and protein content in milk increased, the proportion of Proteobacteria also

increased (table1, P < 0.05). On the contrary, milk protein content (P < 0.05) and SNF

content (P < 0.01) were negatively correlated with Actinobacteria. Milk yield was

positively correlated with Acetobacteraceae and negatively correlated with

Alcaligenaceae. The changes of MUN were significantly correlated with nineteen

family. Among them, eight family such as Staphylococcaceae, Ruminococcaceae and

Moraxellaceae were negatively correlated, while include Enterobacteriaceae,

Pseudomonadaceae and Microbacteriaceae total eleven family were positively

correlated. With the increase of DIM, the contents of Actinobacteria (P < 0.01) and

Prevotellaceae in milk also decreased. With the increase of month age, the proportion

of Acetobacteraceae in milk also increased (table1, P < 0.05), while the proportion of

Caulobacteraceae in milk decreased.

27

Figure 2.4 LEfSe analysis at multiple taxonomic levels of each month data. Cladogram illustrating the taxonomic groups explaining the most variation among

months. Each ring represents a taxonomic level, with phylum (p), class (c), order (o)

and family (f) emanating from the center to the periphery. Each circle is a taxonomic

unit found in the dataset, with circles or nodes shown in colors (other than yellow)

indicating where a taxon was significantly more abundant. Factorial Kruskal-Wallis

test(p<0.05), pairwise Wilcoxon test(p<0.05) and the LDA log-score(value>2.0) data

was marked on this figure. *: p<0.05; **:P<0.01; ***:P<0.001

28

2.4 DISCUSSION

Even with the same feed, the nutrients in milk still showed significant seasonal

changes. Milk in August had the lowest levels of fat and protein throughout the year,

whereas protein and fat content was highest in December (P < 0.05). Although the

month-to-month variations were significant, however, all data did not change over

within the normal range of nutritional criteria (fig1). In the Ralucapavel study, there

was no significant change in nutritional components (Ralucapavel & Gavan, 2011). On

the other hand, there was no monthly change in the number of somatic cells, because

the standard deviation in the number of SCC in the same month was very large,

especially in October milk (n = 37), and the standard deviation in the SCC exceeded

even 50% of the average. This shows that individual differences among cows are

obvious. It is very necessary to monitor individual cows.

Figure 2.5 (A) stacked bar chart of phyla proportion in each month (B) LEfSe

analysis at family levels of each month data. Cladogram illustrating the taxonomic

groups explaining the most variation among months. Each circle is a taxonomic unit

found in the dataset, with circles or nodes shown in colors (other than yellow) indicating

where a taxon was significantly more abundant. Factorial Kruskal-Wallis test(p<0.05),

pairwise Wilcoxon test(p<0.05) and the LDA log-score(value>2.0) data was marked on

this figure. *: p<0.05; **:P<0.01; ***:P<0.001

29

Compared with the stability of milk nutrient composition, the microflora of milk

showed obvious monthly changes and individual differences. Microbiota in milk

samples collected during the relatively warm months of June, August and October were

more than sixty percent similar to each other, while those collected in February and

April were more than sixty percent similar based on S17 Bray-Curtis similarity analysis

by Primer 7 (Data not show). Samples taken in December were different from those

collected in other months. In February, a large proportion of Pseudomonadaceae and

Enterobacteriaceae were found in the milk flora (Fig 4.) It was well-known

psychrotrophic bacteria and are found more often during the winter in raw milk samples

(Hagi et al., 2013; Uraz & Citak, 1998). And some cold-loving Enterobacteriaceae and

Pseudomonadaceae have been found to secrete proteases and lipolytic enzymes that

cause a decline in the sensory properties of cheese (Franciosi et al., 2011). The

proportion of Streptococcaceae and Microbacteriaceae were highest in April than that

in others month. Franciosi found that 45% strains of Streptococcaceae isolates had

obvious thermophilic ability and lipolytic activity. And all Streptococcaceae have

acidifying activity (Franciosi et al., 2011). In June, Enterococcaceae increased

significantly in milk bacteria. Enterococcaceae is in the ascendant of most cheese,

attribute the success to its numerous enzymatic systems, which enable Enterococcaceae

to effectively utilize the nutrients in milk (Montel et al., 2014). In October, the

proportion of Moraxellaceae was the only significant high family compare with others

season. This family are also often found to be associated with milk spoilage (Franciosi

et al., 2011). Staphylococcaceae, also known as a psychrophile, was found in large

quantities in milk collected in December (Franciosi et al., 2011). It can enhance the

flavor, aroma and color of cheese during fermentation (Frétin et al., 2018). These results

suggest that, even in the same farm, the milk microflora produced by different month

may have different effects on the quality of dairy products. These bacterial mainly come

from teat canals and environment (Frétin et al., 2018; Gill et al., 2006)

The total bacterial count (TBC) was positively correlated with somatic cell count

(SCC) (fig 2). it's widely acknowledged that when the number of somatic cells is at a

high level, the shelf life of milk is greatly shortened (Barbano et al., 2006). In this

experiment, TBC should include the TBC1 of milk in the milk breast and TBC2 of

contamination from machines during milking. From the results, we speculate that the

linear correlation between TBC and SCC in Automatic machine collected milk may be

due to the increase of TBC from machines by adhesion to cells. Therefore, in the future,

30

we will further investigate the relationship between the adhesion mechanism of

psychrotrophic bacteria in machines and SCC.

In the selected samples, the quality standards of milk (including TBC and SCC)

were closely related to many bacterial families. When SCC increased, Prevotellaceae

and Bradyrhizobiaceae decreased significantly (P < 0.05). Similarly, TBC was also

negatively correlated with Prevotellaceae. Prevotellaceae, as the dominant bacteria in

intestinal bacteria, was found to be related to grain feeding (Khafipour et al., 2016;

Tang, Han, Yu, Tsuruta, & Nishino, 2017). Meanwhile, TBC was positively correlated

with Enterobacteriaceae (0.341, P < 0.05), which family can lead to disease and

deterioration of milk quality (Bradley, 2002; Franciosi et al., 2011). This indicates that

the proportion of feed may affect the health of dairy cows and the quality of milk.

31

Figure 2.6 The correlation network of different milk components and

predominant family(proportion>1%). Red lines indicate a positive correlation and

Green Line representation has negative correlation. Data was calculated by Spearman’s

rank correlation coefficient with primer 7 and result with significant correlation were

plotted in Cytoscape (Version 3.7.0).

32

On the other hand, the nutrients in milk also can affect the bacteria in milk. When

milk protein and SNF increase, Moraxellaceae in milk will also increase, meanwhile,

Methylobacterium and Bradyrhizobium were deceased. In this experiment,

Moraxellaceae family is mainly composed of Acinetobacter, which would produce

heat-stable extracellular enzymes proteases and lipases to cause the spoilage of milk

(Vithanage et al., 2016). Simultaneously, Propionibacteriaceae in milk are decreasing

while the SNF increasing. Furthermore, Propionibacteriaceae was mainly found in all

milk samples by Propionibacterium acnes, which was considered to be a non-specific

immunodefences (Pyörälä, 2002). In the study of J.S. Hogan, they found the incidence

of clinical mastitis was decreased for cow infused intravenously with 4 mg of killed

Propionibacterium acnes compared with no treatment group (Hogan et al., 1994). It is

also reported that Propionibacteriaceae was a very good probiotic, which regulates

intestinal flora and metabolic activity, promotes intestinal peristalsis, prevents enteritis,

and contains potential inhibitory activity of cancer factors (Cousin et al., 2010; Rabah

et al., 2017). Fat content in milk was positively correlated with Xanthomonadaceae in

milk (0.338, p < 0.05). The Xanthomonadaceae were frequently found in cow feed,

grass and silo samples (Doyle et al., 2017a; Tang et al., 2017; Zeineldin et al., 2017)

Urea content (MUN) in milk has always been used as a monitoring index for nitrogen

use efficiency in dairy cattle (Baker et al., 1995). The content of MUN in milk had

obvious seasonal variation, and the content of MUN in December samples was

significantly lower than that in other months (fig1). In B Martin's study, cheeses made

from milk with high urea content, whether firm, pasty or chalky, were lower than those

made from the control group. However, comparing the two kinds of cheese, the content

of PH or Ca will not change due to the increase of urea (Martin et al., 1997). In our data,

we found that there was a significant positive correlation between urea content in milk

and bacteria that reduced cheese sensory properties and accelerated spoilage of milk,

such as Enterobacteriaceae and Pseudomonadaceae (Table 1; p < 0.01) (Franciosi et

al., 2011). Pseudomonadaceae family and would produce heat-stable extracellular

enzymes aprX-lipA, principally pseudomonas secreted proteases and

Enterobacteriaceae secreted lipases (Vithanage et al., 2016). The proteases casein

micelles to casein gather and precipitate to disequilibria milk gelatinize and lipase can

hydrolyze acylglycerol to release fatty acids like unpleasant flavors short-chain fatty

acids (e.g. butyric, caprylic and caproic acids) and lead to soapy taste medium-chain

fatty acids (Chen et al., 2003; Machado et al., 2017). At the same time, with the increase

33

of MUN, the proportion of Staphylococceae, that enhance flavor in cheese production

will decrease (Frétin et al., 2018; Seitz, 1974).

In addition, we carried out the verification test of the cultivation method. The

results showed that the growth rate of bacteria in E. coli culture medium with 20 mg/dl

urea was significantly faster than that in the control group (PBS group). On the contrary,

the proliferation rate of Staphylococcus. aureus in the culture group with 20 mg/dl urea

was lower than that in the control group (PBS). Similarly, we used milk sold in

supermarkets for validation tests. The rate of PH reduction in milk with artificial urea

(20mg/dl) was significantly higher than that in the control group (P < 0.05) (Data not

displayed).

2.5 CONCLUSION

Results from the present study clearly showed the monthly variation of microbiota

in individual milk. Under the same feeding conditions, the composition of milk

produced in different seasons has obvious seasonal variation. And, the bacteria in milk

also have obvious seasonal changes. These changes may affect the sensory perception

of cheese. In addition, changes in flora in milk may be associated with changes in milk

composition (including fat, protein and MUN). These results suggested that the

composition and microflora of individual milk will affect the quality of tank-milk,

whether in the monitoring of dairy products or milk quality. Monitoring individual flora

and components in milk is very important. According to the season, reasonable

improvement of Feed proportioning may play a positive role in improving the quality

of milk.

2.6 REFERENCE

Auldist, M. J., Coats, S., Sutherland, B. J., Mayes, J. J., McDowell, G. H., & Rogers,

G. L. (1996). Effects of somatic cell count and stage of lactation on raw milk

composition and the yield and quality of Cheddar cheese. Journal of Dairy


Baker, L. D., Ferguson, J. D., & Chalupa, W. (1995). Responses in urea and true

protein of milk to different protein feeding schemes for dairy cows. Journal of

Dairy Science, 78(11), 2424–2434. https://doi.org/10.3168/jds.S0022-

0302(95)76871-0

34

Barbano, D. M., Ma, Y., & Santos, M. V. (2006). Influence of raw milk quality on

fluid milk shelf life. Journal of Dairy Science, 89, E15–E19.


Bradley, A. J. (2002). Bovine mastitis: An evolving disease. The Veterinary Journal,

164(2), 116–128. https://doi.org/10.1053/TVJL.2002.0724

Chen, L., Daniel, R. M., & Coolbear, T. (2003). Detection and impact of protease and

lipase activities in milk and milk powders. International Dairy Journal, 13(4),

255–275. https://doi.org/10.1016/S0958-6946(02)00171-1

Cousin, F. J., Mater, D. D. G., Foligne, B., & Jan, G. (2010). Dairy propionibacteria

as human probiotics: A review of recent evidence. Dairy Science & Technology.

https://doi.org/10.1051/dst/2010032

Doyle, C. J., Gleeson, D., O’Toole, P. W., & Cotter, P. D. (2017a). High-throughput

metataxonomic characterization of the raw milk microbiota identifies changes

reflecting lactation stage and storage conditions. International Journal of Food

Microbiology, 255, 1–6. https://doi.org/10.1016/j.ijfoodmicro.2017.05.019

Doyle, C. J., Gleeson, D., O’Toole, P. W., & Cotter, P. D. (2017b). Impacts of

seasonal housing and teat preparation on raw milk microbiota: A high-

throughput sequencing study. Applied and Environmental Microbiology, 83(2).

https://doi.org/10.1128/AEM.02694-16

Franciosi, E., De Sabbata, G., Gardini, F., Cavazza, A., & Poznanski, E. (2011).

Changes in psychrotrophic microbial populations during milk creaming to

produce Grana Trentino cheese. Food Microbiology, 28(1), 43–51.

https://doi.org/10.1016/j.fm.2010.08.003

Frétin, M., Ferlay, A., Verdier-Metz, I., Fournier, F., Montel, M.-C., Farruggia, A., …

Martin, B. (2017). The effects of low-input grazing systems and milk

pasteurisation on the chemical composition, microbial communities, and sensory

properties of uncooked pressed cheeses. International Dairy Journal, 64, 56–67.


Frétin, M., Martin, B., Rifa, E., Isabelle, V. M., Pomiès, D., Ferlay, A., … Delbès, C.

(2018). Bacterial community assembly from cow teat skin to ripened cheeses is

influenced by grazing systems. Scientific Reports, 8(1), 1–11.

https://doi.org/10.1038/s41598-017-18447-y

Gill, J. J., Sabour, P. M., Gong, J., Yu, H., Leslie, K., & Griffiths, M. W. (2006).

Characterization of bacterial populations recovered from the teat canals of

35

lactating dairy and beef cattle by 16S rRNA gene sequence analysis. FEMS

Microbiology Ecology, 56(3), 471–481. https://doi.org/10.1111/j.1574-

6941.2006.00091.x

Hagi, T., Sasaki, K., Aso, H., & Nomura, M. (2013). Adhesive properties of

predominant bacteria in raw cow’s milk to bovine mammary gland epithelial

cells. Folia Microbiologica, 58(6), 515–522. https://doi.org/10.1007/s12223-013-

0240-z

Hogan, J. S., Smith, K. L., Todhunter, D. A., Schoenberger, P. S., Dinsmore, R. P.,

Canttell, M. B., & Gabel, C. S. (1994). Efficacy of dry cow therapy and a

propionibacterium acnes product in herds with low somatic cell count. Journal of

Dairy Science, 77(11), 3331–3337. https://doi.org/10.3168/JDS.S0022-

0302(94)77274-X

Hwang, S. Y., Lee, M. J., & Chiou, P. W. S. (2000). Monitoring nutritional status of

dairy cows in Taiwan using milk protein and milk urea nitrogen. Asian-

Australasian Journal of Animal Sciences. https://doi.org/10.5713/ajas.2000.1667

Jacobs, J. A., & Siegford, J. M. (2012). Invited review: The impact of automatic

milking systems on dairy cow management, behavior, health, and welfare.

Journal of Dairy Science, 95(5), 2227–2247. https://doi.org/10.3168/jds.2011-

4943





Khafipour, E., Li, S., Tun, H. M., Derakhshani, H., Moossavi, S., & Plaizier, J. C.

(2016). Effects of grain feeding on microbiota in the digestive tract of cattle.

Animal Frontiers, 6(2), 13. https://doi.org/10.2527/af.2016-0018

Kim, I. S., Hur, Y. K., Kim, E. J., Ahn, Y.-T., Kim, J. G., Choi, Y.-J., & Huh, C. S.

(2017). Comparative analysis of the microbial communities in raw milk

produced in different regions of Korea. Asian-Australasian Journal of Animal

Sciences, 30(11), 1643–1650. https://doi.org/10.5713/ajas.17.0689

Latorre, A. A., Van Kessel, J. S., Karns, J. S., Zurakowski, M. J., Pradhan, A. K.,

Boor, K. J., … Sasahara, K. C. (2010). Biofilm in milking equipment on a dairy

farm as a potential source of bulk tank milk contamination with Listeria

monocytogenes. Journal of Dairy Science, 93(6), 2792–2802.

36

https://doi.org/10.3168/jds.2009-2717

Leitner, G., Silanikove, N., Jacobi, S., Weisblit, L., Bernstein, S., & Merin, U. (2008).

The influence of storage on the farm and in dairy silos on milk quality for cheese

production. International Dairy Journal, 18(2), 109–113.


Li, N., Wang, Y., You, C., Ren, J., Chen, W., Zheng, H., & Liu, Z. (2018). Variation

in raw milk microbiota throughout 12 months and the impact of weather

conditions. Scientific Reports, 8(1), 1–10. https://doi.org/10.1038/s41598-018-

20862-8

Machado, S. G., Baglinière, F., Marchand, S., Van Coillie, E., Vanetti, M. C. D., De

Block, J., & Heyndrickx, M. (2017). The biodiversity of the microbiota

producing heat-resistant enzymes responsible for spoilage in processed bovine

milk and dairy products. Frontiers in Microbiology, 8, 302.

https://doi.org/10.3389/FMICB.2017.00302

Martin, B., Coulon, J. B., Chamba, J. F., & Bugaud, C. (1997). Effect of milk urea

content on characteristics of matured Reblochon cheeses. Lait, 77, 505–514.

https://doi.org/10.1051/lait:1997436

Meng, L., Zhang, Y., Liu, H., Zhao, S., Wang, J., & Zheng, N. (2017).

Characterization of Pseudomonas spp. and associated proteolytic properties in

raw milk stored at low temperatures. Frontiers in Microbiology, 8, 1–7.

https://doi.org/10.3389/fmicb.2017.02158

Montel, M. C., Buchin, S., Mallet, A., Delbes-Paus, C., Vuitton, D. A., Desmasures,

N., & Berthier, F. (2014). Traditional cheeses: Rich and diverse microbiota with

associated benefits. International Journal of Food Microbiology, 177, 136–154.

https://doi.org/10.1016/j.ijfoodmicro.2014.02.019

Ni, K., Minh, T. T., Tu, T. T. M., Tsuruta, T., Pang, H., & Nishino, N. (2017).

Comparative microbiota assessment of wilted Italian ryegrass, whole crop corn,

and wilted alfalfa silage using denaturing gradient gel electrophoresis and next-

generation sequencing. Applied Microbiology and Biotechnology, 101(4), 1385–

1394. https://doi.org/10.1007/s00253-016-7900-2

Nousiainen, J., Shingfield, K. J., & Huhtanen, P. (2004). Evaluation of milk urea

nitrogen as a diagnostic of protein feeding. Journal of Dairy Science, 87(2), 386–


Pyörälä, S. (2002). New strategies to prevent mastitis. Reproduction in Domestic

37

Animals, 37(4), 211–216. https://doi.org/10.1046/j.1439-0531.2002.00378.x

Quigley, L., McCarthy, R., O’Sullivan, O., Beresford, T. P., Fitzgerald, G. F., Ross,

R. P., … Cotter, P. D. (2013). The microbial content of raw and pasteurized cow

milk as determined by molecular approaches. Journal of Dairy Science, 96(8),

4928–4937. https://doi.org/10.3168/jds.2013-6688

Rabah, H., Rosa do Carmo, F., & Jan, G. (2017). Dairy propionibacteria: Versatile

probiotics. Microorganisms, 5(2), 24.

https://doi.org/10.3390/microorganisms5020024

Ralucapavel, E., & Gavan, C. (2011). Seasonal and milking-to-milking variations in

cow milk fat, protein and somatic cell counts. Notulae Scientia Biologicae, 3(2),

20–23. http://dx.doi.org/10.15835/nsb325715

Ribeiro Júnior, J. C., de Oliveira, A. M., Silva, F. de G., Tamanini, R., de Oliveira, A.

L. M., & Beloti, V. (2017). The main spoilage-related psychrotrophic bacteria in

refrigerated raw milk. Journal of Dairy Science, 101(1), 75–83.

https://doi.org/10.3168/jds.2017-13069

Rodrigues, M. X., Lima, S. F., Canniatti-Brazaca, S. G., & Bicalho, R. C. (2017). The

microbiome of bulk tank milk: Characterization and associations with somatic

cell count and bacterial count. Journal of Dairy Science, 100(4), 2536–2552.

https://doi.org/10.3168/jds.2016-11540

Seitz, E. W. (1974). Industrial application of microbial lipases: A review. Journal of

the American Oil Chemists’ Society, 51(2), 12–16.

https://doi.org/10.1007/BF02545206

Tang, M. T., Han, H., Yu, Z., Tsuruta, T., & Nishino, N. (2017). Variability, stability,

and resilience of fecal microbiota in dairy cows fed whole crop corn silage.

Applied Microbiology and Biotechnology, 101(16), 6355–6364.

https://doi.org/10.1007/s00253-017-8348-8

Uraz, G., & Citak, S. (1998). The isolation of Pseudomonas and other Gram (-)

psychrotrophic bacteria in raw milks. Journal of Basic Microbiology, 38(2), 129–

134. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/9637013

Verdier-Metz, I., Coulon, J.-B., & Pradel, P. (2001). Relationship between milk fat

and protein contents and cheese yield. Anim. Res, 50, 365.

https://doi.org/10.1051/animres:2001138



38


potential. International Dairy Journal, 57(July), 80–90.


Wouters, J. T., Ayad, E. H., Hugenholtz, J., & Smit, G. (2002). Microbes from raw

milk for fermented dairy products. International Dairy Journal, 12(2–3), 91–109.

https://doi.org/10.1016/S0958-6946(01)00151-0

Yoon, J. T., Lee, J. H., Kim, C. K., Chung, Y. C., & Kim, C. H. (2004). Effects of

milk production, season, parity and lactation period on variations of milk urea

nitrogen concentration and milk components of Holstein dairy cows. Asian-

Australasian Journal of Animal Sciences, 17(4), 479–484.

https://doi.org/10.5713/ajas.2004.479

Zeineldin, M., Lowe, J., de Godoy, M., Maradiaga, N., Ramirez, C., Ghanem, M., …

Aldridge, B. (2017). Disparity in the nasopharyngeal microbiota between healthy

cattle on feed, at entry processing and with respiratory disease. Veterinary

Microbiology, 208, 30–37. https://doi.org/10.1016/j.vetmic.2017.07.006



challenge by high-throughput sequencing, J Sci Food Agric., 95(5):1072-1079.

https://doi.org/10.1002/jsfa.6800

39

CHAPTER 3 Assessment of milk microbiota of dairy cows

managed by automatic milking systems

3.1 INTRODUCTION

The automatic milking system (AMS) was introduced in 1992 and has become

increasingly popular in countries that operate high productivity dairy farming. Without

human participation, an AMS can collect milk at any time when the cows become ready

for milking. The frequency of milking in a day can exceed three times; hence, in

addition to the increase in milk yield, the risk of udder contamination associated with

the farmer’s contact is believed to be minimized (Jacobs & Siegford, 2012). However,

the recovery of the mammary gland could be retarded by the curtailed milking interval,

and infectious mastitis, if present, could potentially spread because the sequence of

milking is not controlled. Therefore, the (Berglund & Pettersson, 2002; Dufour et al.,

2011; Kruip et al., 2002)

Understanding of the milk microbiota is helpful for mastitis prevention (Olde

Riekerink et al., 2007) and food quality management (Jayarao et al., 2004; Quigley et

al., 2013; Vithanage et al., 2016) If mastitis occurs with infectious pathogens, e.g.

Staphylococcus aureus, Streptococcus agalactiae, Corynebacterium bovis, or

Mycoplasma spp., the farmer should reconsider the procedure and sequence of milking.

If mastitis is caused by environmental pathogens, e.g. coliforms, coagulase negative

Staphylococci, or Streptococci other than S. agalactiae, the farmer may take greater

care to improve the hygiene of the cowshed. An understanding of causative

microorganisms helps to decide the appropriate antibiotics for clinical mastitis.

The analysis of milk microbiota is also important for improving the shelf life of

milk. A number of psychrophilic bacteria, e.g. Pseudomonas spp. and Acinetobacter

spp., can produce proteases and lipases even after pasteurization, leading to

organoleptic changes associated with spoilage (Hantsis-Zacharov & Halpern, 2007;

Ribeiro Júnior et al., 2017; Vithanage et al., 2016). The select growth of psychrophilic

bacteria during refrigeration in a bulk cooler is difficult to avoid; however,

contamination through water and the environment needs to be minimized.

A lot of research has been conducted to analyze milk microbiota by both plate-

40

culture and culture-independent methods. Milk microbiota has been shown to be

influenced by many factors including region, season, cowshed environment, and the

hygienic management of milking (Doyle et al., 2017; Elmoslemany et al., 2010; Kable

et al., 2016; Kim et al., 2017). However, the results of next-generation sequencing

(NGS) analyses were highly variable between experiments and the understanding of

high-throughput data is still largely complicated regarding mastitis prevention and milk

quality management (Bhatt et al., 2012; Bonsaglia et al., 2017; Doyle et al., 2017;

Falentin et al., 2016a; Kuehn et al., 2013a; Oikonomou et al., 2014; Young et al., 2015).

Likewise, although the term “raw milk” is often used, the majority of published research

has examined the microbiota of milk stored in cooling tanks (Frétin et al., 2018; Kable

et al., 2016; Kim et al., 2017; Li et al., 2018; Quigley et al., 2013; Rodrigues et al., 2017;

Zhang et al., 2015). Monitoring the microbiota in the tank milk is indeed helpful, but

the data may not be identical to those for udder milk, because of the effect of cooling

(Ogier et al., 2004). Regardless, little information is available on the milk microbiota

of cows managed by AMS.

The advantages of 16S rRNA amplicon sequencing by NGS can be further

improved by incorporating predicted metagenomes using Phylogenetic Investigation of

Communities by Reconstruction (PICRUSt) (Langille et al., 2013). In this study,

microbiota analysis by MiSeq followed with PICRUSt was performed for the milk

collected manually and automatically from the cows managed by an AMS. The aim

was to obtain information on the milk microbiota of the AMS-managed cows and the

effect of the collection procedure on the diversity and predicted gene functions of milk

microbiota.


3.2.1 Sample collection

From approximately 40 lactating Holstein cows raised in the Okayama Prefectural

Livestock Research Institute, three cows were randomly selected. All cows were

managed by AMS (Lely Astronout A4, Cornes AG. Ltd., Eniwa, Japan) with exclusive

feeding of total mixed ration silage throughout the year. The contents of dry matter

(DM), crude protein (N × 6.25), and total digestible nutrients in the total mixed ration

silage were 500–600 g/kg, 160–180 g/kg DM, and 720–740 g/kg DM, respectively.

Milk samples were collected from the same cows in May and July. Manual collection

41

was made between 10:00-11:00, and samples from four quarters were mixed to obtain

a composite sample. Following surface cleaning, several streams of foremilk were

discarded prior to sample collection. Automatic collection was made through an AMS

device into sterile sample cups with preservative (bronopol; 2-bromo-2-nitropropane-

1,3-diol). Samples from the latest visit to the AMS were taken as counterparts of the

manually collected samples. The difference in the time of collection between the

manual and automatic samplings was <6 h. All samples were immediately refrigerated

on ice and transported to Okayama University. Milk yield was recorded from the AMS

samples, and protein, fat, solid-not-fat, and the somatic cell count in milk was

determined using a CombiFoss FT+ (Foss Alle, Hillerød, Denmark).


Bacterial pellets were obtained by centrifugation at 8000 × g for 15 min, and then

washed once with 1 mL of solution containing 0.05 M D-glucose, 0.025 M Tris-HCl

(pH 8.0), and 0.01 M sodium EDTA (pH 8.0). After centrifugation at 15,000 × g for 2

min, bacterial cells were lysed with 180 μL of lysozyme solution (20 g/L lysozyme,

0.02 M Tris-HCl [pH 8.0], 0.002 M sodium EDTA [pH 8.0], 1.2 g/L Triton X-100) at

37°C for 1 h. Subsequent bacterial DNA purification was performed using a

commercial kit (DNeasy Blood & Tissue Kit; Qiagen, Germantown, MD, USA)

according to the manufacturer’s recommendations (Ni et al., 2017).

3.2.3 Quantitative real-time PCR

The total bacterial count was determined by quantitative PCR (qPCR). 2 μL of

each sample DNA solution was added to 23 μL of a PCR mixture containing 12.5 μL

of KAPA SYBR FAST Master Mix (Kapa Biosystems, Inc., Wilmington, MA, USA)

and 8 μM primers targeting the V3 region of the 16S rRNA genes (forward: 5ʹ-ACG

GGGGGCCTACGGGAGGCAGCAG-3′; reverse: 5ʹ-ATTACCGCGGCTGCTGG-

3ʹ). qPCR was performed with a Mini Opticom real time PCR system (Bio-Rad

Laboratories Inc., Tokyo, Japan), with initiation at 95°C for 30 s followed by 35 cycles

of 15 s at 95°C, 20 s at 60°C, and 30 s at 72°C. A standard curve was prepared from

plasmid DNA employing 16S rRNA genes of Escherichia coli (JCM 1649). The copy

number of the standard plasmid was calculated using the molecular weight of the

nucleic acid and the length (base pairs) of the cloned plasmid.

42




CMGCCGCGGTAA-3′; reverse: 5′-GTGACTGGAGTTCAGACGTGTGCTCTT

CCGATCTGGACTACHVGGGTWTCTAAT-3′ ) was employed(Tang, Han, Yu,

Tsuruta, & Nishino, 2017). The PCR protocol was as follows: initiation at 94°C for 2

min and followed by 25 cycles of 94°C for 30 s, 50°C for 30 s, 72°C for 30 s, and a

final elongation of 72°C for 5 min. The products were purified using Fast Gene

Gel/PCR Extraction Kit (NIPPON Genetics Co., Ltd., Tokyo, Japan) and moved to a

second round of PCR with adapter-attached primers. The second PCR protocol was as

follows: initiation at 94°C for 2 min followed by 10 cycles of 94°C for 30 s, 59°C for

30 s, 72°C for 30 s and a final elongation of 72°C for 5 min. The PCR products were

again purified as described above. The purified DNA was then ligated to the 16S rRNA

amplicons prior to 250-bp paired-end sequencing performed on an Illumina MiSeq

platform at FASMAC Co., Ltd. (Kanagawa, Japan).


box microbial ecology pipeline. Before pair-end joining, raw sequence data were

sheared using “sickle pe” to obtain a phred quality score above 30 and ensure that

sequences were longer than 135 bp. Paired-end sequences were joined using fastq-join

with more than 20 bp overlap required between all paired sequences. Chimeric

sequences were identified with USEARCH and removed. The remaining DNA

sequences were grouped into OTU with 97% matched with the closed-reference OTU

picking method in QIIME, with default settings. Both chimera checking and OTU

picking used Greengenes 13.8 as the reference database. Sequences were aligned using

PyNAST. The OTU table was used in the bioinformatics tool PICRUSt (version 1.1.3)

for functional metagenomics. Functional predictions were made using the KEGG

Orthology Pathways. Within PICRUSt, the 16S rRNA copy number was normalized,

molecular functions were predicted, and all results were summarized into KEGG

pathways.




43

Tokyo, Japan) to examine the differences between months and sampling procedures.

Data for microbiota and predicted gene functions were used to perform principle

coordinate analysis and heat map construction using Primer version 7 with Permanova+

add-on software (Primer-E, Plymouth Marine Laboratory, Plymouth, UK). Alpha

diversity indices, i.e. Chao1, Simpson, and Shannon indices, were analyzed by QIIME.

Table 3.1. Milk yield and composition of dairy cows and total bacterial population,

total sequence reads, and alpha diversity indices of microbiota of manually and

automatically collected milk sampled in May and July.

May July

Manual Automatic Manual Automatic

Milk yield (kg/day) 38.7 ± 9.50 40.8 ± 11.2

Milk composition

Protein (%) 3.13 ± 0.41 3.06 ± 0.14

Fat (%) 4.01 ± 0.26 3.27 ± 0.60

Solid-not-fat (%) 8.84 ± 0.52 8.69 ± 0.18

Somatic cell count

(X 103 cells/mL) 189 ± 151 220 ± 148

Milk microbiota

Total bacterial population

(log10 copies/mL) 6.57 ± 0.26 b 7.48 ± 0.71 a 6.93 ± 0.58 b 7.93 ± 1.71 a

Sequence reads 72231 ± 3512 b 96602 ± 7766 a 85483 ± 325 ab 98225 ± 7959 a

Chao 1 index 1649 ± 18 b 1391 ± 133 b 2330 ± 207 a 1361 ± 135 b

Simpson index 0.927 ± 0.01 a 0.883 ± 0.03 a 0.575 ± 0.09 b 0.855 ± 0.01 a

Shannon index 5.96 ± 0.29 a 4.21 ± 0.21 b 3.78 ± 0.56 b 3.95 ± 0.14 b

Values for milk microbiota in the same row with different following superscripted letters are

significantly different (P<0.05).

3.3 RESULTS

Milk yields on the day of sampling were 32.4–49.6 and 32.8–53.6 kg/day in May

and July, respectively. No differences were seen in protein, fat, solid-not-fat contents

in milk between months, although the values appeared lower in July compared with

May. The somatic cell counts (SCC) were 82–362×103 and 50–315×103 cells/mL for

May and July samples, respectively. The cow showing the highest milk yield had an

SCC of >300×103 cells/mL regardless of the month; hence, although there were no

outward signs, the cow could be regarded as having sub-clinical mastitis.

44

Table 3.2. Relative abundances of bacterial families in the microbiota of manually

and automatically collected milk sampled in May and July from dairy cows

managed by automatic milking systems.

May July


Actinomycetaceae 0.53 ± 0.48 0.00 ± 0.00 0.09 ± 0.03 0.01 ± 0.00

Aerococcaceae 11.2 ± 10.4 0.20 ± 0.19 5.29 ± 2.51 0.16 ± 0.16

Bacillaceae 0.08 ± 0.05 0.01 ± 0.01 0.51 ± 0.40 0.07 ± 0.02

Bacteroidaceae 6.44 ± 2.34 a 0.11 ± 0.14 b 1.23 ± 0.48 b 0.29 ± 0.15 b

Clostridiaceae 4.61 ± 2.57 a 0.08 ± 0.09 b 0.59 ± 0.20 ab 0.01 ± 0.00 b

Corynebacteriaceae 4.33 ± 1.41 a 0.09 ± 0.08 b 2.79 ± 1.33 ab 0.16 ± 0.08 b

Enterobacteriaceae 0.74 ± 0.77 c 4.29 ± 0.98 c 73.6 ± 5.72 a 50.4 ± 3.46 b

Enterococcaceae 0.10 ± 0.04 9.95 ± 6.17 0.06 ± 0.02 4.78 ± 1.13

Lachnospiraceae 3.01 ± 0.93 a 0.04 ± 0.04 b 0.73 ± 0.03 b 0.02 ± 0.01 b

Lactobacillaceae 0.54 ± 0.36 0.01 ± 0.01 0.10 ± 0.03 0.03 ± 0.01

Methylobacteriaceae 17.8 ± 2.1 a 0.32 ± 0.36 b 0.01 ± 0.01 b 0.00 ± 0.00 b

Microbacteriaceae 0.04 ± 0.03 0.50 ± 0.45 0.04 ± 0.00 0.04 ± 0.03

Micrococcaceae 0.54 ± 0.09 b 4.25 ± 1.59 a 0.24 ± 0.09 b 2.22 ± 0.78 ab

[Mogibacteriaceae] 1.00 ± 0.36 a 0.01 ± 0.02 b 0.23 ± 0.05 b 0.00 ± 0.00 b

Moraxellaceae 0.86 ± 0.19 b 45.1 ± 10.4 a 0.60 ± 0.26 b 34.1 ± 3.74 a

Peptostreptococcaceae 2.11 ± 1.49 0.04 ± 0.05 0.12 ± 0.03 0.00 ± 0.00

Pseudomonadaceae 0.20 ± 0.07 b 22.5 ± 8.98 a 0.13 ± 0.05 b 6.01 ± 0.84 b

Ruminococcaceae 5.17 ± 1.29 a 0.09 ± 0.10 b 3.82 ± 0.98 a 0.07 ± 0.04 b

Sphingomonadaceae 14.9 ± 5.57 a 0.34 ± 0.43 b 0.01 ± 0.00 b 0.01 ± 0.00 b

Staphylococcaceae 12.4 ± 4.05 a 0.26 ± 0.13 b 2.29 ± 0.93 b 0.29 ± 0.12 b

Streptococcaceae 1.57 ± 1.38 b 11.2 ± 4.43 a 0.87 ± 0.65 b 0.40 ± 0.23 b

[Tissierellaceae] 1.37 ± 0.35 a 0.02 ± 0.03 c 0.68 ± 0.12 b 0.24 ± 0.16 bc

Turicibacteraceae 1.25 ± 0.96 0.01 ± 0.01 0.08 ± 0.01 0.00 ± 0.00

Others 9.25 ± 2.22 a 0.46 ± 0.38 b 5.83 ± 0.88 a 0.66 ± 0.31 b

The families detected at >1% of the total in at least one sample are presented. Values in the same

row with different following superscripted letters are significantly different (P<0.05).

Although milk samples were obtained from the same cows in May and July, large

differences were seen in the microbiota between the two sampling months.

Methylobacteriaceae (17.8%), Sphingomonadaceae (14.9%), Staphylococcaceae

(12.4%), Aerococcaceae (11.2%), and Bacteroidaceae (6.4%) were the five most

abundant families in samples manually collected in May. In contrast, an overwhelming

abundance of Enterobacteriaceae (73.6%) and a low abundance of Aerococcaceae

(5.3%), Ruminococcaceae (3.8%), Staphylococcaceae (2.3%), and Bacteroidaceae

(1.2%) were seen in samples manually collected in July.

Substantial differences of milk microbiota were demonstrated between manually

collected and automatically collected samples. Moraxellaceae (45.1%),

45

Pseudomonadaceae (22.5%), Streptococcaceae (11.2%), Enterobacteriaceae (4.3%),

and Micrococcaceae (4.2%) were predominant in samples automatically collected in

May, and Enterobacteriaceae (50.4%), Moraxellaceae (34.1%), Pseudomonadaceae

(6.0%), Enterococcaceae (4.8%), and Micrococcaceae (2.2%) were major families in

samples automatically collected in July. Regarding the five most abundant families in

the May-collected samples, no taxa were found in common between manually collected

and automatically collected samples. Likewise, only Enterobacteriaceae was found as

a common predominant taxon between samples manually collected and automatically

collected in July. Accordingly, Principal Coordinates Analysis (PCoA) at the 60%

similarity level indicated a clear separation of milk microbiota due to months and

sampling procedures (Fig 2a). At the 80% similarity level, samples collected in May

were separated almost individually both for manually collected and automatically

collected milk.

Total bacterial populations were greater for automatically collected than manually

collected samples, whereas no differences were seen between the May and July samples.

Data for manually collected samples indicated that the Simpson and Shannon indexes

were greater and Chao 1 was lower in May than in the July-collected samples.

There were 6908 KEGG orthology groups aligned with normalized OTU data, and

55 second level pathways from eight first level pathways contributed to the description

of gene functions and metabolic activities. The comparison of manually collected

samples with criteria of differences between months at >0.5% indicated that amino acid

metabolism, energy metabolism, nucleotide metabolism, replication and repair,

translation, xenobiotics biodegradation and metabolism were greater in the May-

collected samples, whereas glycan biosynthesis and metabolism and membrane

transport were higher in the July-collected samples. For automatically collected

samples, amino acid metabolism, replication and repair, translation, and xenobiotics

biodegradation and metabolism were greater in the May-collected samples, and

membrane transport was higher in the July-collected samples. Unlike manually

collected samples, energy metabolism and nucleotide metabolism were similar between

May and July, but lipid metabolism and metabolism of terpenoids and polyketides were

greater in May for automatically collected samples. The PCoA of the predicted gene

functions demonstrated that, regardless of 60 and 80% similarity levels, all milk

samples were regarded as belonging to the same group.

46

Figure 3.1 Association heatmap between the relative abundance of bacterial

families, sampling procedure, and sampling month for the milk of cows managed

by automatic milking systems. The families detected at >1% of the total in at least

one milk sample are presented. M and A stand for manually collected and automatically

collected sample, respectively.

47

Figure 3.2 Principal coordinates analyses of (a) bacterial families and (b) KEGG

pathways in the microbiota of manually and automatically collected milk sampled

in May and July from dairy cows managed by automatic milking systems. The

families and pathways detected at >1% of the total in at least one milk sample are

presented. M and A stand for manually collected and automatically collected sample,

respectively.

48

Figure 3.3 The KEGG pathways encoded in the microbiota of (a) manually

collected and (b) automatically collected milk. Only the KEGG pathways

significantly affected by sampling month are presented. Positive values indicate an

increased representation in May compared with July.

3.4 DISCUSSION

In this study, healthy cows with no outward signs of clinical mastitis were used to

examine milk microbiota for two months and for two sampling procedures. Differences

between May and July were quite large especially for the manually collected samples,

and Enterobacteriaceae exceeded 70% in the samples that were manually collected

July. Li et al. (Li et al., 2018) reported a large variation in the microbiota of tank milk

according to the months and seasons, and the relative abundance of Proteobacteria was

about 60% in July. Enterobacteriaceae is regarded as a gut-associated bacterial group,

and several genera like Escherichia coli, Klebsiella spp., and Proteus spp. are known

to be involved in environmental mastitis (Makovec & Ruegg, 2003). High temperature

49

and humidity in summer might have enhanced their growth in faeces and bedding.

However, Kagkli et al. indicated that the ribotypes of coliform bacteria isolated from

milk were not the same as those isolated from bovine faeces; hence, a large proportion

of Enterobacteriaceae detected in July samples may not be originated from faeces and

bedding (Kagkli et al., 2007). Relative abundance of Enterobacteriaceae could increase

to 60% in the milk of cows with mastitis (Bhatt et al., 2012); however, the abundance

itself would not help detect a cow with mastitis, because even the cow showing SCC of

<50×103 cells/mL had Enterobacteriaceae at 60% abundance in this study.

Figure 3.4 The KEGG pathways encoded in the microbiota of manually collected

and automatically collected milk. Only the KEGG pathways significantly affected by

sampling method are presented. Positive values indicate an increased representation in

Manual compared with AMS milk.

Methylobacteriaceae, Sphingomonadaceae, and Staphylococcaceae were

abundant in samples manually collected in May. Methylobacteriaceae are oligotrophic

bacteria and, although several reports indicated their presence in bovine milk (Bracke

50

et al., 2014), they are regarded as a minor group and unrelated to mastitis and milk

quality. Sphingomonas spp. have the ability to grow in a wide range of environments

that are not tolerated by most other bacteria. Although Kuehn et al. reported that, based

on NGS analysis, the relative abundance of Sphingomonas spp. was 20.4% in the

mastitis-affected quarter and 4.0% in the healthy quarter, Sphingomonas spp. are not

regarded as mastitis-causing bacteria (Kuehn et al., 2013b). Staphylococcus aureus is a

representative agent of infectious mastitis (Makovec & Ruegg, 2003), and Bhatt et al.

detected as much as 50% of Staphylococcus spp. in milk from cows with mastitis (Bhatt

et al., 2012).

Aerococcaceae was most abundant (25.8%) in one sample manually collected in

May. Aerococcus spp. can be detected in milk as the predominant bacteria (Quigley et

al., 2013), and faeces is regarded as the source of Aerococcus spp. Although A. viridians

is known as a mastitis-associated pathogen (Liu et al., 2015), the cow that showed the

highest abundance of Aerococcaceae in this study had an SCC of 82×103 cells/mL.

Thus, a high abundance of Aerococcus spp. does not necessarily indicate sub-clinical

and clinical mastitis.

There were no distinctive differences in diet composition between May- and July-

collected samples and, although protein, fat, and solid-not-fat contents appeared lower

in July, no significant differences were seen in milk composition and SCC between

May and July in this study. However, a large variation was demonstrated in milk

microbiota, especially when manually collected milk was used for assessment. Cow-

to-cow variation was small compared with month-to-month variation in milk

microbiota; hence, a group of cows managed in the same environment displayed similar

monthly variations of milk microbiota.

51

Table 3.3. The KEGG pathways encoded in the microbiota of manually and automatically collected milk sampled in May and July from dairy cows managed

by automatic milking systems.

May July


Amino acid metabolism 9.76 ± 0.20 b 10.61 ± 0.28 a 8.80 ± 0.02 c 9.27 ± 0.05 bc

Carbohydrate metabolism 10.5 ± 0.08 ab 9.71 ± 0.58 b 10.2 ± 0.04 ab 10.9 ± 0.09 a

Cell motility 2.99 ± 0.31 a 2.05 ± 0.36 bc 2.43 ± 0.07 ab 1.61 ± 0.04 c

Energy metabolism 6.42 ± 0.09 a 5.97 ± 0.13 b 5.95 ± 0.03 b 5.76 ± 0.02 b

Enzyme families 1.95 ± 0.02 b 1.75 ± 0.02 c 2.02 ± 0.00 a 1.95 ± 0.01 b

Folding, sorting and degradation 2.31 ± 0.02 ab 2.40 ± 0.08 a 2.31 ± 0.01 ab 2.19 ± 0.02 b

Function unknown 1.52 ± 0.08 c 1.97 ± 0.09 b 2.14 ± 0.04 ab 2.20 ± 0.01 a

General function prediction only 3.65 ± 0.02 a 3.59 ± 0.01 b 3.49 ± 0.01 c 3.33 ± 0.01 d

Glycan biosynthesis and metabolism 1.83 ± 0.10 c 1.98 ± 0.05 c 2.55 ± 0.05 a 2.30 ± 0.01 b

Lipid metabolism 3.25 ± 0.10 b 3.94 ± 0.18 a 3.04 ± 0.00 b 3.31 ± 0.04 b

Membrane transport 11.5 ± 1.07 b 12.8 ± 0.70 b 15.8 ± 0.11 a 17.0 ± 0.19 a

Metabolism of cofactors and vitamins 4.64 ± 0.18 a 4.19 ± 0.06 b 4.23 ± 0.01 b 4.13 ± 0.01 b

Metabolism of other amino acids 1.79 ± 0.08 bc 1.96 ± 0.05 a 1.63 ± 0.00 c 1.79 ± 0.01 b

Metabolism of terpenoids and polyketides 1.86 ± 0.04 b 2.18 ± 0.08 a 1.51 ± 0.02 c 1.63 ± 0.03 c

Nucleotide metabolism 3.87 ± 0.15 a 3.38 ± 0.17 bc 3.45 ± 0.05 b 3.05 ± 0.01 c

Other ion-coupled transporters 1.34 ± 0.08 c 1.60 ± 0.04 b 1.70 ± 0.01 b 1.87 ± 0.01 a

Replication and repair 8.21 ± 0.30 a 6.99 ± 0.18 b 7.01 ± 0.12 b 5.97 ± 0.02 c

Signal transduction 2.19 ± 0.17 b 2.26 ± 0.15 ab 2.58 ± 0.05 a 2.55 ± 0.00 ab

Transcription 2.40 ± 0.12 b 2.40 ± 0.09 b 2.88 ± 0.01 a 2.98 ± 0.03 a

Translation 5.09 ± 0.29 a 4.44 ± 0.13 b 4.13 ± 0.10 b 3.56 ± 0.02 c

Xenobiotics biodegradation and metabolism 2.67 ± 0.12 b 3.82 ± 0.24 a 2.10 ± 0.02 c 2.76 ± 0.07 b

Others 10.3 ± 0.47 10.0 ± 0.15 10.1 ± 0.10 9.91 ± 0.03

The pathways detected at >1% of the total in at least one milk sample are presented. Values in the same row with different following superscripted letters are significantly

different (P<0.05).

52

In addition to the major families detected in this study (Aerococcaceae,

Enterobacteriaceae, Methylobacteriaceae, Moraxellaceae, Pseudomonadaceae,

Sphingomonadaceae, Staphylococcaceae, and Streptococcaceae), NGS analysis has

demonstrated that Bacillaceae (Li et al., 2018), Bifidobacteriaceae (Frétin et al., 2018),

Corynebacteriaceae (Bonsaglia et al., 2017), Flavobacteriaceae (Zhang et al., 2015),

Lachnospiraceae (Falentin et al., 2016b), Ruminococcaceae (Rodrigues et al., 2017),

and Propionibacteriaceae (Oikonomou, Bicalho, Meira, Rossi, Foditsch, Machado,

Gustavo, et al., 2014) could be detected at >10% abundance in raw milk. Bacillaceae,

Bifidobacteriaceae, Flavobacteriaceae, and Propionibacteriaceae were always <1.0%,

but Ruminococcaceae, Corynebacteriaceae, and Lachnospiraceae exceeded 3.0% in

several samples regardless of the sampling month in this study. Alpha diversity data

indicated that a great number of bacterial taxa were dispersed in samples manually

collected in May, whereas an extreme deviation of Enterobacteriaceae was evident in

samples manually collected in July. This distinctive difference in diversity between

May- and July-collected samples was not clear when microbiota was assessed with

automatically collected milk. Likewise, the five most abundant families were

completely different between samples manually collected (Methylobacteriaceae,

Sphingomonadaceae, Staphylococcaceae, Aerococcaceae, and Bacteroidaceae) and

automatically collected (Moraxellaceae, Pseudomonadaceae, Streptococcaceae,

Enterobacteriaceae, and Micrococcaceae) in May. Further, Moraxellaceae,

Pseudomonadaceae, Enterococcaceae, and Enterobacteriaceae were the four most

abundant families regardless of sampling months in automatically collected samples.

Therefore, although milk microbiota varied greatly because of changing environment

and other undefined factors, the variation may not be clearly seen when assessed with

automatically collected milk. In this study, the time differences between manual and

automatic samplings were <6 h. For psychrophilic bacterial proliferation in milk, 6 h

may be too short. Therefore, major families (Moraxellaceae and Pseudomonadaceae)

stably found in automatically collected samples were probably derived from teat cups,

pipe lines, and hoses in the AMS device.

Although microbiota was substantially different between May and July and

between manually and automatically collected samples, gene functions predicted by

PICRUSt were mostly unchanged and the extent of the differences was rather small.

Regardless of sampling procedures, the activities of amino acid metabolism, replication

53

and repair, translation, and xenobiotics biodegradation and metabolism were higher in

May, and the activity of membrane transport was greater in July. An overwhelming

proportion of Enterobacteriaceae in July-collected samples was probably involved in

enhanced membrane transport, but the relationship between Enterobacteriaceae

proportion and membrane transport function is difficult to explain.

Despite the fact that the microbiota was substantially different between manually

collected samples from May and July, differences in gene functions were <2%, except

for membrane transport, which was about 5%. This finding suggested that dysbiosis of

microbiota would not imply disorganization of gene functions to the same extent. The

fact that all four milk samples grouped together by gene function, but clearly separated

by microbiota supports this.

Figure 3.5 The predicted functions based on FAPROTAX database in the

microbiota of manually collected and automatically collected milk. Only the KEGG

pathways significantly affected by sampling method are presented. Positive values

indicate an increased representation in Manual compared with AMS milk.

54

The distinctive differences in gene functions between samples manually collected

May and July were in the categories of energy metabolism, glycan biosynthesis and

metabolism, and nucleotide metabolism, and those between samples automatically

collected May and July were lipid metabolism, metabolism of terpenoids and

polyketides, and transcription. The meaning of these differences is not clear, but the

emphasized lipid metabolism in automatically collected milk could be related to the

abundance of psychrophilic Moraxellaceae and Pseudomonadaceae bacterial species.

3.5 CONCLUSION

The milk microbiota of healthy dairy cows managed by AMS was examined.

Large differences were seen between samples manually collected May and July; the

total abundance of Methylobacteriaceae, Sphingomonadaceae, and Staphylococcaceae

bacteria was about half as large in May-collected samples compared to July-collected

samples, but an overwhelming abundance (>70%) of Enterobacteriaceae was seen in

July-collected samples. The microbiota of automatically collected milk was

substantially different from manually collected milk; Moraxellaceae,

Pseudomonadaceae, and Enterococcaceae were major families present in both May-

and July-collected samples. Although the milk microbiota greatly differed between the

May and July samples, the differences in predicted gene functions were <2%, except

for membrane transport. Although milk microbiota can vary greatly between months in

cows managed by AMS, the variation may be less pronounced when milk samples are

collected automatically compared with manually.

3.6 REFERENCES

Berglund, I., & Pettersson, G. (2002). Automatic milking: Effects on somatic cell

count and teat end-quality. Livestock Production Science, 78(2), 115–124.

https://doi.org/10.1016/S0301-6226(02)00090-8

Bhatt, V. D., Ahir, V. B., Koringa, P. G., Jakhesara, S. J., Rank, D. N., Nauriyal, D.

S., … Joshi, C. G. (2012). Milk microbiome signatures of subclinical mastitis-

55

affected cattle analysed by shotgun sequencing. Journal of Applied

Microbiology, 112(4), 639–650. https://doi.org/10.1111/j.1365-

2672.2012.05244.x

Bonsaglia, E. C. R., Gomes, M. S., Canisso, I. F., Zhou, Z., Lima, S. F., Rall, V. L.

M., … Lima, F. S. (2017). Milk microbiome and bacterial load following dry

cow therapy without antibiotics in dairy cows with healthy mammary gland.

Scientific Reports, 7(1), 1–10. https://doi.org/10.1038/s41598-017-08790-5

Bracke, N., Van Poucke, M., Baert, B., Wynendaele, E., De Bels, L., Den Broeck, W.

Van, … De Spiegeleer, B. (2014). Identification of a microscopically selected

microorganism in milk samples. Journal of Dairy Science, 97(2), 609–615.

https://doi.org/10.3168/jds.2013-6932

Doyle, C. J., Gleeson, D., O’Toole, P. W., & Cotter, P. D. (2017). Impacts of seasonal

housing and teat preparation on raw milk microbiota: A high-throughput

sequencing study. Applied and Environmental Microbiology, 83(2).

https://doi.org/10.1128/AEM.02694-16

Dufour, S., Fréchette, A., Barkema, H. W., Mussell, A., & Scholl, D. T. (2011).

Invited review: Effect of udder health management practices on herd somatic cell

count. Journal of Dairy Science, 94(2), 563–579.

https://doi.org/10.3168/jds.2010-3715

Elmoslemany, A. M., Keefe, G. P., Dohoo, I. R., Wichtel, J. J., Stryhn, H., &

Dingwell, R. T. (2010). The association between bulk tank milk analysis for raw

milk quality and on-farm management practices. Preventive Veterinary

Medicine, 95(1–2), 32–40. https://doi.org/10.1016/j.prevetmed.2010.03.007

Falentin, H., Rault, L., Nicolas, A., Bouchard, D. S., Lassalas, J., Lamberton, P., …

Even, S. (2016a). Bovine teat microbiome analysis revealed reduced alpha

diversity and significant changes in taxonomic profiles in quarters with a history

of mastitis. Front Microbiol, 7:480. https://doi.org/10.3389/fmicb.2016.00480

Frétin, M., Martin, B., Rifa, E., Isabelle, V. M., Pomiès, D., Ferlay, A., … Delbès, C.

(2018). Bacterial community assembly from cow teat skin to ripened cheeses is

influenced by grazing systems. Scientific Reports, 8(1), 1–11.

https://doi.org/10.1038/s41598-017-18447-y

Hantsis-Zacharov, E., & Halpern, M. (2007). Culturable psychrotrophic bacterial

communities in raw milk and their proteolytic and lipolytic traits. Applied and

Environmental Microbiology, 73(22), 7162–7168.

56

https://doi.org/10.1128/AEM.00866-07

Jacobs, J. A., & Siegford, J. M. (2012). Invited review: The impact of automatic



4943

Jayarao, B. M., Pillai, S. R., Sawant, A. A., Wolfgang, D. R., & Hegde, N. V. (2004).

Guidelines for monitoring bulk tank milk somatic cell and bacterial counts.


0302(04)73493-1





Kagkli, D. M., Vancanneyt, M., Vandamme, P., Hill, C., & Cogan, T. M. (2007).

Contamination of milk by enterococci and coliforms from bovine faeces. Journal

of Applied Microbiology, 103(5), 1393–1405. https://doi.org/10.1111/j.1365-

2672.2007.03338.x









Phillips, G. J. (2013b). Bacterial community profiling of milk samples as a

means to understand culture-negative bovine clinical mastitis. PLoS ONE, 8(4).


Langille, M. G. I., Zaneveld, J., Caporaso, J. G., Mcdonald, D., Knights, D., Reyes, J.

A., … Huttenhower, C. (2013). Analysis predictive functional profiling of

microbial communities using 16S rRNA marker gene sequences. Nature

Biotechnology, 31(9), 814–821. https://doi.org/10.1038/nbt.2676


in Raw Milk Microbiota Throughout 12 Months and the Impact of Weather

57

Conditions. Scientific Reports, 8(1), 1–10. https://doi.org/10.1038/s41598-018-

20862-8

Liu, G., Liu, Y., Ali, T., Ferreri, M., Gao, J., Chen, W., … Han, B. (2015). Molecular

and phenotypic characterization of Aerococcus viridans associated with

subclinical bovine mastitis. PLoS ONE, 10(4), 1–14.


Makovec, J. A., & Ruegg, P. L. (2003). Results of Milk Samples Submitted for

Microbiological Examination in Wisconsin from 1994 to 2001. Journal of Dairy

Science, 86(11), 3466–3472. https://doi.org/10.3168/jds.S0022-0302(03)73951-4





1394. https://doi.org/10.1007/s00253-016-7900-2

Ogier, J., Girard, V., Leveau, J., & Gruss, A. (2004). Raw Cow Milk Bacterial

Population Shifts Attributable to Refrigeration, 70(9), 5644–5650.

https://doi.org/10.1128/AEM.70.9.5644


S., … Bicalho, R. C. (2014). Microbiota of cow’s milk; distinguishing healthy,

sub-clinically and clinically diseased quarters. PLoS ONE, 9(1).



S., … Bicalho, R. C. (2014). Microbiota of Cow’ s Milk; Distinguishing Healthy,

Sub-Clinically and Clinically Diseased Quarters, 9(1).


Olde Riekerink, R. G. M., Barkema, H. W., & Stryhn, H. (2007). The effect of season

on somatic cell count and the incidence of clinical mastitis. Journal of Dairy


Quigley, L., McCarthy, R., O’Sullivan, O., Beresford, T. P., Fitzgerald, G. F., Ross,



4928–4937. https://doi.org/10.3168/jds.2013-6688

Quigley, L., O’Sullivan, O., Stanton, C., Beresford, T. P., Ross, R. P., Fitzgerald, G.

F., & Cotter, P. D. (2013). The complex microbiota of raw milk. FEMS

58

Microbiology Reviews, 37(5), 664–698. https://doi.org/10.1111/1574-

6976.12030

Ribeiro Júnior, J. C., de Oliveira, A. M., Silva, F. de G., Tamanini, R., de Oliveira, A.

L. M., & Beloti, V. (2017). The main spoilage-related psychrotrophic bacteria in

refrigerated raw milk. Journal of Dairy Science, 101(1), 75–83.

https://doi.org/10.3168/jds.2017-13069




https://doi.org/10.3168/jds.2016-11540




https://doi.org/10.1007/s00253-017-8348-8




potential. International Dairy Journal, 57(July), 80–90.


Young, W., Hine, B. C., Wallace, O. A. M., Callaghan, M., & Bibiloni, R. (2015).

Transfer of intestinal bacterial components to mammary secretions in the cow.

PeerJ, 3, e888. https://doi.org/10.7717/peerj.888



challenge by high-throughput sequencing. Journal of the Science of Food and

Agriculture, 95(5), 1072–1079. https://doi.org/10.1002/jsfa.6800

59

CHAPTER 4 Rumen fluid, feces, milk, water, feed, airborne

dust, and bedding microbiota in dairy farms managed by

automatic milking systems

4.1 INTRODUCTION

Mastitis, an inflammation of the mammary gland regarded as the most important

disease affecting dairy herds, is triggered by pathogens derived from infectious and

environmental bacteria. Assessment of milk microbiota is thus important for preventing

mastitis and maintaining herd health (Jayarao et al., 2004; Olde Riekerink et al., 2007).

Typical infectious bacteria include Staphylococcus aureus, Streptococcus agalactiae,

Corynebacterium bovis, and Mycoplasma spp. Therefore, if mastitis is caused by these

pathogens the farmer should revise their milking procedure and sequence. If

environmental bacteria, such as coliforms, coagulase negative Staphylococci, and

Streptococci other than S. agalactiae, are pathogenic agents, the farmer should improve

the hygiene of their cowshed.

Implementation of automatic milking systems (AMS) is one of the most important

technological advancements in the dairy industry in the past 20 years. The use of AMS

enables cows to be milked whenever they want. Therefore, milk production may

increase as the milking frequency of individual cow increases as compared with regular

twice per day milking (Kruip et al., 2002). Moreover, milk quality, cow behavior, cow

welfare, and herd management have been shown to be affected by AMS (Jacobs &

Siegford, 2012) and the risk of udder contamination associated with farmer contact is

expected to be minimized. However, the teat orifice could be damaged more when using

an AMS than when employing conventional milking techniques, because the curtailed

milking interval may retard the recovery of the mammary gland. Further, infectious

mastitis, if present, could spread throughout the herd via the AMS, because the

sequence of milking is not controlled. Though, studies have shown inconsistent results

regarding the risk of mastitis during the transition from conventional milking to AMS

(Berglund & Pettersson, 2002; Dufour et al., 2011; Jacobs & Siegford, 2012).

A lot of research has been conducted, by both plate-culture and culture-

independent methods, to analyze milk microbiota in association with milking practices

and the farm management. Milk microbiota is shown to be influenced by the microbiota

60

present on teat skin, bedding, feed (hay), and in the surrounding air (Doyle et al., 2017;

Quigley et al., 2013; Vacheyrou et al., 2011). Likewise, region, season, cowshed

environment, and hygiene of the milking practices are known environmental factors

that influence the microbiota (Elmoslemany et al., 2010; Kable et al., 2016; Kim et al.,

2017).

Although increasing data for milk microbiota, determined by next-generation

sequencing (NGS) analyses, have become available, the results were highly variable

between experiments and the understanding of high-throughput data is largely

complicated by factors such as milk quality control and mastitis prevention (Bhatt et

al., 2012; Bonsaglia et al., 2017; Doyle et al., 2017; Falentin et al., 2016; Kuehn et al.,

2013; Oikonomou et al., 2014; Young et al., 2015). Further, information on the milk

microbiota of cows managed by AMS is lacking.

In this study, microbiota analysis by NGS (MiSeq) was performed for rumen

fluid, feces, milk, water, feed (total mixed ration silage), bedding, and airborne dust

collected at two dairy farms managed by AMS. The aim of this study was to

characterize the microbiota of the gut, milk, and cowshed environment. Diet and

nutrition were shown to affect the composition of milk microbiota and the risk of

mastitis (Zhang et al., 2015). Therefore, feed, rumen fluid, and feces were examined to

see how diet and gut microbiota associate with each other and raw milk.



Samples were collected at two farms at Hiroshima and Okayama Japan, located

at >100 km away from each other, which operated AMS (Lely Astronout A4, Cornes

AG. Ltd., Eniwa, Japan) for management of lactating dairy cows. At both farms the

cows were housed in a free stall barn and fed total mixed ration silage, which was

formulated to have 500–600 g/kg of dry matter (DM), 160–180 g/kg DM of crude

protein (N×6.25), and 720–740 g/kg DM of total digestible nutrients. Samples were

collected between 10:00–12:00 in July at farm 1 and between 13:00–15:00 in October

at farm 2. Rumen fluid was obtained using a flexible stainless spring tube (Lumenar

stomach evacuator outfit, Fujihira Industry Co. Ltd., Tokyo, Japan), and fecal samples

were collected from the rectum. Milk samples were collected manually from four

udders and then mixed as a composite sample. Following surface cleaning, several

61

streams of foremilk were discarded prior to sample collection. Airborne dust samples

were collected by placing three petri dishes for five minutes approximately 1.0 m above

the ground. Bedding samples were collected from three separate places in a cowshed

and water was collected from three different water cups. Feed was sampled by taking

three piles of total mixed ration silage. In the free stall system, cows could move and

rest freely and determining their resting place was difficult. Thus, a composite sample

prepared from three separate samples was thus regarded as a representative means of

assessing airborne dust, water, bedding, and feed microbiota at a farm. All samples

were immediately refrigerated on ice and transported to Okayama University.

Procedures and protocols for the animal experiments were approved by the Animal

Care and Use Committee, Okayama University, Japan.


Bacterial DNA from feed, milk(250µL), water(250µL), and airborne dust(250µL)

samples was extracted and purified using the DNeasy Blood & Tissue Kit (Qiagen,

Germantown, MD, USA). Bacterial pellets were obtained by centrifugation at 16,000

× g for 2 min. For feed samples, 10 g of silage was vigorously mixed with 90 mL of

sterilized saline and the gauze-filtered extract was centrifuged to obtain pellets (Ni et

al., 2016). All pellet samples were lysed with 180 µL of lysozyme solution (20 g/L

lysozyme, 0.02 M Tris-HCl [pH 8.0]), 0.002 M sodium EDTA [pH 8.0], 1.2 g/L Triton

X-100) at 37°C for 1 hr. Subsequent bacterial DNA purification was performed

following the manufacturer’s recommendations. For rumen fluid, feces, and bedding

samples, 0.1 g of the sample was used to prepare bacterial pellets and bacterial DNA

was purified using the DNeasy Stool Mini Kit (Qiagen, Germantown, MD, USA).

4.2.3 Quantitative real-time PCR for total bacteria

The total bacterial count was determined by quantitative PCR (qPCR). Each

sample DNA solution (2 μL) was added to 23 μL of a PCR mixture containing 12.5 μL


and 8 μM primers targeting the V3 region of the 16S rRNA genes (forward: 5ʹ-

ACGGGGGGCCTACGGGAGGCAGCAG-3′; reverse: 5ʹ-ATTACCGCGGCTGCTG

G-3ʹ). The qPCR was performed with a Mini Opticon real time PCR system (Bio-Rad

Laboratories Inc., Tokyo, Japan) with initiation at 95°C for 30 sec followed by 35 cycles

62

of 15 sec at 95°C, 20 sec at 60°C, and 30 sec at 72°C. A standard curve was prepared

from plasmid DNA employing 16S rRNA genes from Escherichia coli (JCM 1649).

The copy number of the standard plasmid was calculated using the molecular weight of

the nucleic acid and the length (base pairs) of the cloned plasmid.




CMGCCGCGGTAA-3′; reverse: 5′-GTGACTGGAGTTCAGACGTGTGCTCTTCC

GATCTGGACTACHVGGGTWTCTAAT-3′) was employed (Tang, Han, Yu, Tsuruta,

& Nishino, 2017). The PCR protocol was as follows: initiation at 94°C for 2 min and

followed by 25 cycles of 94°C for 30 sec, 50°C for 30 sec, 72°C for 30 sec, and a final

elongation of 72°C for 5 min. The products were purified using the Fast Gene Gel/PCR

Extraction Kit (NIPPON Genetics Co., Ltd., Tokyo, Japan) and moved to a second

round of PCR with adapter-attached primers. The second PCR protocol was as follows:

initiation at 94°C for 2 min followed by 10 cycles of 94°C for 30 sec, 59°C for 30 sec,

72°C for 30 sec and a final elongation of 72°C for 5 min. The PCR products were again

purified as described above. The purified DNA was then ligated to the 16S rRNA

amplicons prior to 250 bp paired-end sequencing performed on an Illumina MiSeq

platform at FASMAC Co., Ltd. (Kanagawa, Japan).

4.2.5 Bioinformatics and microbiota characterization







sequences were grouped into an OTU with 97% matched with the closed-reference

OTU picking method in QIIME, when assessed with default settings. Both chimera

checking and OTU picking used Greengenes 13.8 as the reference database and

sequences were aligned using PyNAST. The results of the sequence analysis are

available in the DDBJ Sequence Read Archive under project identification number

63

PRJDB7427.

4.2.6 Statistical analyses

Comparison of total population and bacterial composition between the two farms

was examined by analysis of variance. Microbiota data were also subjected to canonical

analysis of principal coordinates to define assignment and clustering that explained

variations in the microbiota. Discriminant vectors with a Pearson correlation >0.7 were

considered significant. Likewise, hierarchical clustering and heat map construction

were done. These analyses were performed using Primer version 7 with Permanova+

add-on software (Primer-E, Plymouth Marine Laboratory, Plymouth, UK).

Sources of environmental contamination of milk were assessed using the

SourceTracker algorithm (Knights et al., 2011). During this source tracking, airborne

dust, bedding, water, and feed microbiota were regarded to be a common source of

contamination on a farm, i.e. the source of milk contamination could vary between cows

even if they were kept in a same housing.

4.3 RESULTS

Regardless of the farms, Prevotellaceae (31.9 and 25.5% respectively at farm 1

and 2) was the most abundant taxa in rumen fluid microbiota (Fig. 1, Fig. S1, and Table

S1). Other bacteria identified included Ruminococcaceae (11.2%), Lachnospiraceae

(9.3%), Paraprevotellaceae (2.9%), and Veillonellaceae (1.7%) at farm 1, and

Succinivibrionaceae (13.3%), Ruminococcaceae (10.8%), Lachnospiraceae (5.2%),

and Veillonellaceae (4.4%) at farm 2.

The most abundant taxa in the feces microbiota were different from those in rumen

fluid microbiota. Ruminococcaceae was found at 38.5 and 39.2% at farm 1 and 2

respectively. Additionally, Lachnospiraceae (7.8%), Clostridiaceae (6.6%),

Bacteroidaceae (6.1%), and Peptostreptcoccaceae (3.0%) were found at farm 1, and

Bacteroidaceae (11.5%), Lachnospiraceae (5.1%), Clostridiaceae (4.5%), and

Rikenellaceae (3.5%) were detected at farm 2.

The five most abundant taxa in the milk microbiota were Aerococcaceae (24.3%),

Staphylococcaceae (12.3%), Ruminococcaceae (11.4%), Corynebacteriaceae (5.9%),

and Lachnospiraceae (5.1%) at farm 1, and Staphylococcaceae (21.0%),

Lactobacillaceae (10.8%), Ruminococcaceae (6.3%), Corynebacteriaceae (6.1%), and

64

Enterobacteriaceae (5.6%) at farm 2. Although Aerococcaceae and Staphylococcaceae

were present in the two highest proportions in the milk microbiota at farm 1 and 2, one

out of three cows had Ruminococcaceae as the most abundant taxa at both farms.

Although only one composite sample of bedding was examined for each farm,

Ruminococcaceae (19.5 and 10.8% at farm 1 and 2 respectively) was found as the

predominant taxa at the two farms. Some taxa varied greatly between farms with

Aerococcaceae (15.0%), Staphylococcaceae (9.7%), Corynebacteriaceae (8.8%), and

Lachnospiraceae (6.4%) observed at farm 1, and Moraxellaceae (10.4%),

Idiomarinaceae (8.5%), Halomonadaceae (8.2%), and Corynebacteriaceae (7.0%)

observed at farm 2.

In airborne dust microbiota, Aerococcaceae (25.2%) was the most abundant at

farm 1 followed by Ruminococcaceae (12.0%), Staphylococcaceae (10.3%),

Lachnospiraceae (5.8%), and Corynebacteriaceae (5.7%). At farm 2, Lactobacillaceae

(64.5%) were found at far greater proportions than Staphylococcaceae (5.6%),

Ruminococcaceae (3.1%), Pseudomonadaceae (2.2%), and Aerococcaceae (1.8%).

Regardless of the farms, Lactobacillaceae (38.8 and 55.7% at farm 1 and 2

respectively) was the most abundant taxa in the water microbiota. Other taxa were seen

at proportions <10% including Comamonadaceae (6.8%), Moraxellaceae (5.7%),

Pseudomonadaceae (5.1%), and Staphylococcaceae (3.6%) at farm 1, and

Moraxellaceae (9.5%), Aeromonadaceae (4.3%), Neisseriaceae (4.2%), and

Weeksellaceae (3.8%) at farm 2.

At both farms, total mixed ration silage was exclusively fed to dairy cows and the

proportions of Lactobacillaceae exceeded 95% in the feed microbiota. The second most

prominent taxon was Leuconostocaceae, but the proportions were as low as 1.0–2.6%.

According to heatmap, the rumen fluid and feces microbiota were clearly separated

with the bedding microbiota (Fig. 2). Few differences were seen in the rumen fluid and

feces microbiota between individual cows across the two farms. Farm-to-farm and cow-

to-cow differences in milk microbiota appeared to be greater compared with the feed,

rumen fluid, and feces microbiota. The milk microbiota at farm 1 were grouped with

the airborne dust and bedding microbiota. Although the milk microbiota for two

samples at farm 2 was grouped with the bedding microbiota, that for one sample was

related with the airborne dust and water microbiota.

65


at >1.0% at least two different samples) in feed, rumen fluid, feces, milk, bedding, water, and

airborne dust microbiota in two dairy farms managed by an automatic milking system. Clustering

was performed using the Euclidean distance as a similarity metric. F1, F2, rumen, and air indicate

farm 1, farm 2, rumen fluid, and airborne dust, respectively

66

Canonical analysis of principal coordinates further clarified the taxa associated

with rumen fluid, feces, milk, water, feed, airborne dust, and bedding samples (Fig. 3).

Rumen fluid and feces, which were characterized by Prevotellaecae and

Ruminococcaceae respectively, were regarded as separate groups at a 60% similarity

level. As depicted in the heatmap, the milk, airborne dust, and bedding at farm 1 were

considered to be in a same group, which was characterized with a high abundance of

Aerococcaceae. Feed, airborne dust, and water at farm 2 formed one group and was

characterized by the high abundance of Lactobacillaceae. Differences with respect to

airborne dust and bedding between farm 1 and 2 were apparent.

Figure 4.2 Family level proportions of feed, rumen fluid, feces, milk, bedding,


milking system. F1, F2, rumen, and air indicate farm 1, farm 2, rumen fluid, and

airborne dust, respectively

67

The SourceTracker algorithm was used to identify the likely source of milk

microbiota in the dairy farm environment (Fig. 3). Regardless of the farms, airborne

dust was identified as the greatest contributor (53.0 and 37.9% at farm 1 and 2,

respectively) to the milk microbiota, followed by feces (13.8%), bedding (13.7%), and

water (4.30%) at farm 1, and bedding (9,70%), feces (8.00%), and rumen fluid (6.40%)

at farm 2. Farm-to-farm differences were not evident (P>0.05) for any source of

contamination. When SourceTracker analysis was performed by combining the data

from the two farms; the contributions of airborne dust, feces, and bedding were

calculated to be 45.5, 10.9, and 10.1%, respectively.

4.4 DISCUSSION

Although the proportion of Prevotellaceae is the most abundant taxa (>25%) in

the rumen fluid, the proportion was <0.3% in feces. Ruminococcaceae was the second

(farm 1) and the third (farm 2) most abundant taxa in the rumen fluid, but the relative

abundance was the highest in feces for any cows regardless of the farm. Prevotellaceae

is regarded as a major soluble carbohydrate degrader and the proportion correlated with

grain feeding (Khafipour et al., 2016). Therefore, differences in the taxon proportions

between rumen fluid and feces indicated that availability of the soluble carbohydrates

was greatly lowered over digestion from the rumen to the large intestine.

Lachnospiraceae was found at similar proportions in rumen fluid and feces, whereas

Veillonellaceae was higher in rumen fluid and Bacteroidaceae and Clostridiaceae were

higher in feces. Ruminococcaceae, Bacteroidaceae, Lachnospiraceae, and

Clostridiaceae were identified as the four most abundant taxa in feces, which was

similar to that seen in our previous study (Tang et al., 2017). The observation that

Prevotellaceae is predominant in rumen fluid whereas the abundance was greatly

different between the rumen fluid and feces was also similar to that reported by Mao

(Mao et al., 2015).

Well-preserved TMR silage with Lactobacillaceae at >95% was exclusively fed

to the cows at both farms. Therefore, the observation that the mean proportion of

Succinivibrionaceae in rumen microbiota was numerically higher at farm 2 (13.3%)

than farm 1 (0.92%) was difficult to explain. However, one cow among three cows at

farm 2 had Succinivibrionaceae at <1.0%, which was similar to the mean proportion of

68

Succinivibrionaceae in the rumen fluid at farm 1. Thus, although Succinivibrionaceae

has been known to correlate with grain feeding (Khafipour et al., 2016), the taxa may

show a large cow-to-cow variation in rumen fluid.

Figure 4.3 Canonical analysis of principal coordinates plot characterizing the

microbiota of feed, rumen fluid, feces, milk, bedding, water, and airborne dust in

the two dairy farms managed by an automatic milking system. The operational

taxonomy unit with Pearson's correlation >0.7 is overlaid on the plot as vectors.

Samples for farm 1 and 2 are presented as red and blue plots, respectively. Samples

enclosed in a green circle are regarded to be in the same group at a 60% similarity level.

F1, F2, rumen, and air indicate farm 1, farm 2, rumen fluid, and airborne dust,

respectively

69

Figure 4.4 Pie charts of the percentages of inferred sources of milk

microbiota in two dairy farms managed by an automatic milking system. Cows

moved freely and determining their resting place was difficult in the free stall system.

A composite sample prepared from three separate samples was thus regarded as a

representative means of assessing airborne dust, water, bedding, and feed microbiota at

a farm. The values are the means and standard deviations for three cows

Staphylococcaceae, Ruminococcaceae, and Corynebacteriaceae were

identified as the most abundant taxa in the milk microbiota at both farms.

Staphylococcus aureus is the representative agent of infectious mastitis and the

proportion of Staphylococcus spp. was shown to increase to >50% in clinical mastitis

(Bhatt et al., 2012). Relative abundances of Staphylococcaceae detected in this study

(12.3% and 21.0% at farm 1 and 2 respectively) should be regarded as high levels, but

the cows from which milk was collected did not show any symptoms of clinical or

subclinical mastitis in this study.

70

Although a lot of studies have examined the milk microbiota of cows with and

without mastitis, no typical microbiota has been defined as the microbiota for healthy

cow’s milk. Bonsaglia reported that Corynebacterium spp. (Corynebacteriaceae) and

Psychrobacter spp. (Moraxellaceae) were the two most abundant taxa in milk samples

and Acinetobacter spp. (Moraxellaceae), Staphylococcus spp (Bonsaglia et al., 2017).

(Staphylococcaceae), and Micrococcus spp. (Micrococcaceae) were found at >5.0%

abundance. Falentin found that Oscillospira spp. (Ruminococcaceae) and

Staphylococcus spp. (Staphylococcaceae) were prevalent taxa and Bifidobacterium spp.

(Bifidobacteriaceae) was found as the next most abundant taxon in the milk microbiota

(Falentin et al., 2016). Because milk microbiota can vary between seasons (Li et al.,

2018), to define the microbiota of healthy cow’s milk may be difficult.

In the water microbiota, Lactobacillaceae and Moraxellaceae were found in both

farm 1 and 2. A high proportion of Lactobacillaceae may indicate a transfer from mouth

to cows fed Lactobacillaceae-rich TMR silage, and that of Moraxellaceae could

indicate the presence of the psychrophilic Acinetobacter spp. The relative abundance

of Lactobacillaceae was high in airborne dust at farm 2. Therefore, a high proportion

of Lactobacillaceae in water may also indicate a transfer from the surrounding air.

At both farm 1 and 2, Ruminococcaceae and Corynebacteriaceae were found as

the major taxa in the bedding microbiota, which agrees with the findings of Doyle

(Doyle et al., 2017). Ruminococcaceae is regarded as a gut inhabitant, whereas

Corynebacteriaceae is known to inhabit diverse environments. Our results indicating

that milk and feces microbiota are separately grouped from bedding microbiota were

similar to those of Doyle (Doyle et al., 2017). Ruminococcaceae and Lachnospiraceae

were identified as the major taxa in feces, bedding, and milk microbiota at farm 1,

indicating that milk microbiota could be contaminated by gut-associated groups

(Oikonomou et al., 2014). However, neither Ruminococcaceae nor Lachnospiraceae

are considered to be mastitis pathogens. Rather, Aerococcaceae, Staphylococcaceae,

and Corynebacteriaceae were found at high relative abundances. Therefore, non-gut

groups like A. viridans, S. aureus, and C. bovis might provoke mastitis at farm 1.

Regardless, at two farms examined in this study, and apparently well managed by AMS,

gut-associated microbiota was not a primary risk factor for mastitis.

71

Figure 4.5 The five most abundant taxa in rumen fluid, milk, feces, bedding,

airborne dust, water, and feed microbiota in two dairy farms managed by an

automatic milking system.

Aerococcaceae, Ruminococcaceae, and Staphylococcaceae were major taxa in the

airborne dust microbiota at farm 1 and 2. Evgrafov reported that the major taxa of

microbiota of airborne dust collected at a milking parlor were Lachnospiraceae (26%),

Aerococcaceae (12.9%), Peptostreptococcaceae (11.0%), and Moraxellaceae (6.2%)

and Staphylococcaceae (2.0%), Corynebacteriaceae (2.5%), Lactobacillaceae (1.1%),

and Pseudomonadaceae (1.7%) were low in relative abundance (De Evgrafov et al.,

2013). Although they reported that these bacteria were detectable in non-farm outdoor

samples collected 8 km away from the dairy farm, they did not detect Ruminococcaceae

in either the cowshed or non-farm outdoor environment. In this study, three airborne

72

dust samples collected at the free-barn were composited. Therefore, Ruminococcaceae

have a greater likelihood of detection than those samples collected near milking parlor.

The observation that Lactobacillaceae is a major taxon in airborne dust at farm 2 was

difficult to explain, but Lactobacillaceae was found at a high relative abundance

(10.8%) in water at this farm. Thus, although the route and source were unclear,

Lactobacillaceae could become a major taxon of the airborne dust microbiota in the

cow shed environment. Luongo et al. (2017) reported that Lactobacillus spp.

(Lactobacillaceae), Streptococcus spp. (Streptococcaceae), Micrococcus spp.

(Micrococcaceae), Corynebacterium spp. (Corynebacteriaceae), Haemophilus spp.

(Pasteurellaceae), and Finegoldia spp. (Peptoniphilaceae) were found as the major

taxa in the airborne dust microbiota of non-farm indoor samples (university dormitory

rooms).

Although relative abundances were different between milk and airborne dust,

Ruminococcaceae, Aerococcaceae, and Staphylococcaceae were the three most

abundant taxa found in common between milk and airborne dust at farm 1. Likewise,

Staphylococcaceae, Lactobacillaceae, and Ruminococcaceae were the three most

abundant taxa found in common between milk and airborne dust at farm 2. Indeed,

SourceTracker indicated that the milk microbiota may be related to the airborne dust

microbiota in non-mastitis healthy cows. Further, Corynebacteriaceae was found at

both farm 1 and 2 at stable relative abundances (5.8―8.8%) in milk and bedding

microbiota. Corynebacteriaceae in milk may have been derived from that in the

bedding. Therefore, management of the bedding should be a primary consideration as

a measure to prevent mastitis.

Based on hierarchical clustering and canonical analysis of principal coordinates,

milk microbiota, particularly at farm 1, was associated with the bedding microbiota.

Although Doyle et al. (2016) clarified that the teat microbiota showed the greatest

similarity with the milk microbiota, we did not examine teat or udder skin microbiota

in this study. The relationship between airborne dust, bedding, and the teat microbiota

should be determined in the forthcoming studies.

4.5 CONCLUSION

This study examining the microbiota of the gut, milk, and cowshed environment

in dairy farms managed by AMS displayed greater farm-to-farm and cow-to-cow

73

differences in milk microbiota compared with the feed, rumen fluid, and feces

microbiota. Milk microbiota appeared to be influenced by airborne dust based on the

source tracking. Hierarchical clustering and canonical analysis of principal coordinates

demonstrated that the milk microbiota was associated with the bedding microbiota but

clearly separated from feed, rumen fluid, feces, and water microbiota. Our findings

were derived from only two case studies at Chugoku Region in Japan. Therefore, it is

unclear if these findings should be limited to the farms managed by AMS. Regardless,

the importance of cowshed management should be emphasized to maintain cow’s

health and prevent mastitis.

4.6 REFERENCES

Berglund, I., & Pettersson, G. (2002). Automatic milking: Effects on somatic cell

count and teat end-quality. Livestock Production Science, 78(2), 115–124.

https://doi.org/10.1016/S0301-6226(02)00090-8





2672.2012.05244.x

Bonsaglia, E. C. R., Gomes, M. S., Canisso, I. F., Zhou, Z., Lima, S. F., Rall, V. L.

M., … Lima, F. S. (2017). Milk microbiome and bacterial load following dry

cow therapy without antibiotics in dairy cows with healthy mammary gland.

Scientific Reports, 7(1), 1–10. https://doi.org/10.1038/s41598-017-08790-5

De Evgrafov, M. R., Kõll, P., Frank, D. N., Baumgartner, L. K., Robertson, C. E.,

Hernández, M. T., & Pace, N. R. (2013). Molecular analysis of bacterial and

circovirus bioaerosols in concentrated animal feeding operations. Aerosol

Science and Technology, 47(7), 755–766.

https://doi.org/10.1080/02786826.2013.789477



sequencing study. Applied and Environmental Microbiology, 83(2), 1–12.

https://doi.org/10.1128/AEM.02694-16

Dufour, S., Fréchette, a, Barkema, H. W., Mussell, a, & Scholl, D. T. (2011). Invited

74

review: effect of udder health management practices on herd somatic cell count.

Journal of Dairy Science, 94(2), 563–579. https://doi.org/10.3168/jds.2010-3715

Elmoslemany, A. M., Keefe, G. P., Dohoo, I. R., Wichtel, J. J., Stryhn, H., &

Dingwell, R. T. (2010). The association between bulk tank milk analysis for raw

milk quality and on-farm management practices. Preventive Veterinary

Medicine, 95(1–2), 32–40. https://doi.org/10.1016/j.prevetmed.2010.03.007

Falentin, H., Rault, L., Nicolas, A., Bouchard, D. S., Lassalas, J., Lamberton, P., …

Even, S. (2016). Bovine teat microbiome analysis revealed reduced alpha

diversity and significant changes in taxonomic profiles in quarters with a history

of mastitis. Frontiers in Microbiology, 7(APR).

https://doi.org/10.3389/fmicb.2016.00480

Jacobs, J. a, & Siegford, J. M. (2012). Invited review: The impact of automatic



4943




0302(04)73493-1





Khafipour, E., Li, S., Tun, H. M., Derakhshani, H., Moossavi, S., & Plaizier, J. C.

(2016). Effects of grain feeding on microbiota in the digestive tract of cattle.

Animal Frontiers, 6(2), 13. https://doi.org/10.2527/af.2016-0018





Knights, D., Kuczynski, J., Charlson, E. S., Zaneveld, J., Mozer, M. C., Collman, R.

G., … Kelley, S. T. (2011). Bayesian community-wide culture-independent

microbial source tracking. Nature Methods, 8(9), 761–765.

https://doi.org/10.1038/nmeth.1650

75





Phillips, G. J. (2013). Bacterial Community Profiling of Milk Samples as a

Means to Understand Culture-Negative Bovine Clinical Mastitis, 8(4).



in Raw Milk Microbiota Throughout 12 Months and the Impact of Weather

Conditions. Scientific Reports, 8(1), 1–10. https://doi.org/10.1038/s41598-018-

20862-8

Mao, S., Zhang, M., Liu, J., & Zhu, W. (2015). Characterising the bacterial

microbiota across the gastrointestinal tracts of dairy cattle: Membership and

potential function. Scientific Reports, 5(April), 1–14.

https://doi.org/10.1038/srep16116


S., … Bicalho, R. C. (2014). Microbiota of cow’ s milk; Distinguishing healthy,

sub-clinically and clinically diseased quarters. PLoS One, 9(1):e85904.


Olde Riekerink, R. G. M., Barkema, H. W., Stryhn, H., Riekerink, R. G. M. O.,

Barkema, H. W., & Stryhn, H. (2007). The effect of season on somatic cell count

and the incidence of clinical mastitis. Journal of Dairy Science, 90(4), 1704–

1715. https://doi.org/10.3168/jds.2006-567

Quigley, L., O’Sullivan, O., Stanton, C., Beresford, T. P., Ross, R. P., Fitzgerald, G.

F., & Cotter, P. D. (2013). The complex microbiota of raw milk. FEMS

Microbiology Reviews, 37(5), 664–698. https://doi.org/10.1111/1574-

6976.12030




https://doi.org/10.1007/s00253-017-8348-8

Vacheyrou, M., Normand, A. C., Guyot, P., Cassagne, C., Piarroux, R., & Bouton, Y.

(2011). Cultivable microbial communities in raw cow milk and potential

transfers from stables of sixteen French farms. International Journal of Food

76

Microbiology, 146(3), 253–262.


Young, W., Hine, B. C., Wallace, O. A. M., Callaghan, M., & Bibiloni, R. (2015).

Transfer of intestinal bacterial components to mammary secretions in the cow.

PeerJ, 3, e888. https://doi.org/10.7717/peerj.888





77

CHAPTER 5 Milk composition and microbiota in dairy

farms in China

5.1 Introduction

Mastitis is one of the most expensive livestock diseases. Infected cows will not be

able to produce milk for nearly a month. At the same time, the feeding costs of dairy

cows, including labor and feed costs, have been continuously consumed. Furthermore,

in order to treat disease costs, including veterinary costs, drug costs, and so on, and

further increase the burden of breeding. As an infectious disease, infected dairy cows

also have the possibility of infecting other dairy cows. Some cattle farms choose to

eliminate sick cows under multiple considerations (Jayarao et al, 2004; Ma et al., 2016).

Therefore, finding the law of mastitis infection will bring great benefits to milk

production.

Generally, the main cause of mastitis is the invasion of standard pathogens,

including Escherichia coli, Staphylococcus, Streptococcus and so on, into the

mammary glands of dairy cows (Patel et al., 2017). As the main source of bacteria in

milk, the bacteria in the environment will directly affect the bacterial flora in milk.

When the feeding mode changes, the microflora in milk produced by indoor feeding

and grazing cows will change significantly (Doyle et al., 2017). As the main feeding

method of modern dairy cows, the bacteria in milk produced by indoor feeding cows

mainly come from dust bacteria in the air and bacteria on the ground (Wu et al., 2019).

Therefore, a reasonable method of dairy cattle feeding and management will obviously

affect the bacteria in milk produced by dairy cattle in the dairy farm. It is also important

to understand whether the flora in milk of cows’ mammary glands in different areas are

identical or similar under the same or similar feeding and management methods, which

will play a decisive role in the formulation of cow farm management.

As a nutritious food, milk is an excellent hotbed for bacteria. The contents of milk,

including protein, fat, lactose and so on, will provide the nutrition needed for the large-

scale reproduction of bacteria in milk. As we know, when the nutrients in the culture

medium change, the bacterial flora will also change significantly. With the development

of the times, the current sequencing of 16srRNA gene for bacteria MiSeq can clearly

analyze the proportion of bacteria in the sample (Kable et al., 2016; Li et al., 2018).

78

The total number of bacteria can be collected by QPCR. By calculating, the absolute

number of bacteria in each sample will be presented to us. Combining the nutritional

components in milk and the total bacterial count in milk, it may provide beneficial data

support for the prevention of pathogenic bacteria of cow mastitis. Therefore, we plan

to investigate the bacterial flora in milk of individual dairy cows from different cattle

farms in different regions under similar management conditions to see if the same

management is significantly different due to geographical differences. In addition, no

matter which cattle farm, the relationship between nutritional components in milk and

bacterial flora in milk will be analyzed.

5.2Materials and methods


From five dairy farms implementing milking parlor and free barn housing in

Heilongjiang and Shanxi provinces, China. All dairy cows were fed with commercial

silage. Milk is collected manually and in accordance with the collection procedure.

Before the dairy cows were required to produce milk, the on-duty veterinarians in the

dairy hall used alcohol cotton to clean the nipples and manually collected and discarded

the milk that had just been squeezed under 3-4 times. About 15 ml of milk was collected

separately from the breasts of randomly selected five dairy cows in each farm. The

collected milk is kept in ice until it is analyzed. Milk protein, fat, solid-not-fat, lactose, MUN

and the somatic cell count in milk was determined using a CombiFoss FT+ (Foss Alle, Hillerød,

Denmark).


A total of 25 samples from three farms in Heilongjiang Province and two farms in

Shanxi Province were used for DNA extraction. Bacterial pellets were obtained by






79


Qiagen, Germantown, MD, USA) according to the manufacturer’s recommendations

(Ni et al., 2017).

5.2.3 Quantitative real-time PCR

The total bacterial count was determined by quantitative PCR (qPCR). 2 μL of

each sample DNA solution was added to 23 μL of a PCR mixture containing 12.5 μL


and 8 μM primers targeting the V3 region of the 16S rRNA genes (forward: 5ʹ-

ACGGGGGGCCTACGGGAGGCAGCAG-3′; reverse: 5ʹ-

ATTACCGCGGCTGCTGG-3ʹ). qPCR was performed with a Mini Opticon real time

PCR system (Bio-Rad Laboratories Inc., Tokyo, Japan), with initiation at 95°C for 30

s followed by 35 cycles of 15 s at 95°C, 20 s at 60°C, and 30 s at 72°C. A standard

curve was prepared from plasmid DNA employing 16S rRNA genes of Escherichia coli

(JCM 1649). The copy number of the standard plasmid was calculated using the

molecular weight of the nucleic acid and the length (base pairs) of the cloned plasmid.




CMGCCGCGGTAA-3′; reverse:5′-GTGACTGGAGTTCAGACGTGTGCTCTTCCG

ATCTGGACTACHVGGGTWTCTAAT-3′) was employed (Tang, Han, Yu, Tsuruta,


followed by 25 cycles of 94°C for 30 s, 50°C for 30 s, 72°C for 30 s, and a final

elongation of 72°C for 5 min. The products were purified using Fast Gene Gel/PCR



initiation at 94°C for 2 min followed by 10 cycles of 94°C for 30 s, 59°C for 30 s, 72°C

for 30 s and a final elongation of 72°C for 5 min. The PCR products were again purified

as described above. The purified DNA was then ligated to the 16S rRNA amplicons

prior to 250-bp paired-end sequencing performed on an Illumina MiSeq platform at

80

FASMAC Co., Ltd. (Kanagawa, Japan).










PyNAST.




Tokyo, Japan) to examine the differences between farms and regions. Data for

microbiota and predicted gene functions were used to perform principle coordinate

analysis and heat map construction using Primer version 7 with Permanova+ add-on

software (Primer-E, Plymouth Marine Laboratory, Plymouth, UK). Alpha diversity

indices, i.e. Evenness, observed otus, and Shannon indices, were rarefied to an equal

depth of 30,000 sequences by QIIME. LEfSe Analysis of Seasonal Change by

Galaxy(http://huttenhower.sph.harvard.edu/galaxy). The correlation between the milk



5.3 Results

The fat content, lactose content and urea nitrogen content in milk have obvious

differences between farm and farm. However, there are no obvious regional differences.

In addition, the total bacterial count and somatic cell count in milk did not show

significant differences between cattle farms and between regions. The milk of

Heilongjiang cattle farm B had the highest fat content, the largest variation of somatic

http://huttenhower.sph.harvard.edu/galaxy

81

cell number and the most bacterial species.

A

B

Figure 5.1 Farm variation of milk composition(A) and alpha diversity(B) by one-

way ANOVA. Heilongjiang farm A(HA), farm B (HB) and farm C(HC), and Shanxi

farm A(SA) and farm B(SB) was abbreviated

Pathogenic bacteria of mastitis, including Enterobacteriaceae, Staphylococcaceae

and streptococcaceae, can be found in milk of individual cows. However, the

emergence of pathogenic bacteria does not have obvious changes between farm and

farm, between regions and regions. In addition, pseudomonadaceae and moraxellaceae

often appear in the milk of individual cows, and no obvious changes between regions

or between cattle farms have been found.

82


at >1.0% at least two different samples) in milk microbiota in five dairy farms

collected from Heilongjiang and Shanxi. Clustering was performed using the

Euclidean distance as a similarity metric. Heilongjiang farm A(HA), farm B (HB) and

farm C(HC), and Shanxi farm A(SA) and farm B(SB) was abbreviated

83


manually collected milk sampled from different farm in China. The families and

pathways detected at >1% of the total in at least one milk sample are presented.

Heilongjiang farm A(HA), farm B (HB) and farm C(HC), and Shanxi farm A(SA) and

farm B(SB) was abbreviated

There are obvious individual differences between dairy cows in the same cattle

farm. When the Enterobacteriaceae in milk became the dominant bacterial sample, the

bacterial flora phase velocity in milk was higher than 60% regardless of region to region

or farm to farm. Similarly, when Staphylococcaceae is becoming the dominant bacteria

in milk, the similarity of bacteria in milk is higher than 40%. No obvious regional

differences were found.

In lefse analysis, milk produced by Heilongjiang cattle farm and Shanxi cattle farm

had higher proportion of bacteria, such as Bacteroidetes phylum, Cyanobacteria

84

phylum Chloroplast class, Cytophagia class, Anaerolineae class, Streptophyta ordor,

GMD14H09 ordor, Cyrophagaordor family and parapretellaceae family in Shanxi

milk than Heilongjiang milk. Only Yaniellaceae family in Heilongjiang milk samples

was significantly higher than that in Shanxi milk samples.

Figure 2.4 LEfSe analysis at family levels of Heilongjiang and Shanxi in China.

Each circle is a taxonomic unit found in the dataset, with circles or nodes shown in

colors indicating where a taxon was significantly more abundant. Factorial Kruskal-

Wallis test(p<0.05), pairwise Wilcoxon test(p<0.05) and the LDA log-score(value>2.0)

SCC in milk was positively correlated with fat content, protein content and total

bacterial count (TBC). Lachnospiraceae, Planococcaceae, Ruminococcaceae,

Hyphomicrobiaceae, Alteromonadaceae, Intrasporangiaceae and Micrococcaceae in

milk were positively correlated. Actinobacteria, Proteobacteria, Firmicutes and

Bacteroidetes in milk increase with the increase of total bacteria in milk, regardless of

region or cattle farm. The protein content in milk was positively correlated with

Enterobacteriaceae, Ruminococcaceae, Aerococcaceae, Planococcaceae and

Lachnospiraceae. The fat content in milk was correlated with Planococcaceae,

Hyphomicrobiaceae and Pseudomonadaceae. The TBC in milk increased with the

increase of pathogenic-related family such as Streptococcaceae, Enterococcaceae and

Staphylococcaceae.

85

Figure 5.5 The correlation network of different milk components and

predominant family(proportion>1%). Red lines indicate a positive correlation and

Green Line representation has negative correlation. Data was calculated by Spearman’s

rank correlation coefficient with primer 7 and result with significant correlation were

plotted in Cytoscape (Version 3.7.0).

86

5.4 Discussion

In this experiment, the two areas were nearly 1,000 kilometers away and all the

samples were collected manually. The ambient temperature of Heilongjiang cattle farm

was 22-33 degrees, and that of Shanxi cattle farm was 16-25 degrees. Meanwhile, the

sampling days in Shanxi Province were rainy and cloudy while those in Heilongjiang

Province were sunny. Under such a huge environmental difference, only a few non-

dominant bacteria have obvious regional changes, and the main bacterial changes are

mainly individual differences. Although some individual dairy cows have higher SCC

content in their milk, because they do not show any symptoms associated with mastitis,

all dairy cows are certified as healthy by veterinarians on the dairy farm. The fat content

of milk produced in Heilongjiang B cattle farm is obviously higher than that in other

cattle farms, and the individual difference of somatic cell number is the greatest. Further

analysis also showed that there was a significant positive correlation between fat

content and somatic cell number in milk. This may indicate that fat content in milk may

stimulate mammary epithelial cells by some means, leading to a significant increase in

SCC in milk. Gargouri suggested that the increase of somatic cell count in milk might

be related to lipase in milk (Gargouri et al., 2008). In this study, fat content and SCC

content in milk were positively correlated with Planococcaceae and

Hyphomicrobiaceae. Although little research has been done on the ability of these two

bacteria to cause diseases, data analysis shows that these bacteria may be directly or

indirectly related to the increase of SCC.

When the standard pathogenic bacteria, Enterobacteriaceae and

Staphylococcaceae in milk showed a large proportion (more than 5%), the bacterial

flora in milk tended to be similar in regardless of farm to farm or region to region. It

can be found that the protein content in milk is positively correlated with the percentage

of Enterobacteriaceae in the relationship between nutrients and bacteria (Zhang et al.,

2015). Similarly, in the second chapter, we also found that urea nitrogen in milk was

positively correlated with Enterobacteriaceae. This may indicate that the proportion of

Enterobacteriaceae in milk will increase significantly when the protein digestion

balance of cows’ changes (Bi et al., 2016). Increased proportion of Enterobacteriaceae

may lead to mastitis (Blum et al., 2017). On the other hand, from the introduction of

the first chapter, we can know that the incidence of Enterobacteriaceae mastitis in

summer milk is significantly higher than that in other seasons. In previous studies, we

87

found that this may be related to the fact that Enterobacteriaceae in the air caused by

the continuous use of fans in cattle farms in summer is more likely to contaminate the

breast of dairy cows than in other seasons (Kable et al., 2016; Laura et al., 2009). In

this season, cows with abnormal protein digestion are prone to Enterobacteriaceae

mastitis (Wu et al., 2019).

Unlike environmental pollution, Ruminococcaceae in milk has a significant

positive correlation with SCC, but it is difficult to enter contaminated milk in vitro.

This is because Ruminococcaceae is more frequently found in rumen samples and cow

feces samples. And as an anaerobic bacterium, Ruminococcaceae family is very

difficult to survive in the aerobic environment in vitro. Now, some studies are further

exploring how intestinal bacteria can transfer bacteria through in vivo mechanisms.

This correlation was also found in Cornell's laboratory studies (Rodrigues et al., 2017).

5.5 Conclusion

In this study, bacterial flora in milk from five dairy farms, which were produced

in different areas but managed similarly, were investigated. Under similar management

conditions, the regional differences of bacterial flora in cow milk were not obvious. The

change of dominant bacteria is mainly the individual difference between cattle and

cattle. When the content of pathogenic bacteria family increased significantly, the

characteristic flora changes appeared in milk. The nutritional content of milk has

obvious correlation with bacteria in milk. The level of nutrient secretion in milk of dairy

cows may significantly affect the microflora in milk, thus leading to the occurrence of

diseases in dairy cows.

5.6 Reference

Bi, Y., Wang, Y. J., Qin, Y., Vallverdú, R. G., & García, J. M. (2016). Prevalence of

bovine mastitis pathogens in bulk tank milk in China. PLoS One,

11(5):e0155621. https://doi.org/10.1371/journal.pone.0155621

Blum, S. E., Heller, E. D., Jacoby, S., Krifucks, O., & Leitner, G. (2017). Comparison

of the immune responses associated with experimental bovine mastitis caused by

different strains of Escherichia coli. Journal of Dairy Research, 84(2), 190–197.

88

https://doi.org/10.1017/S0022029917000206




https://doi.org/10.1128/AEM.02694-16

Gargouri, A., Hamed, H., & ElFeki, A. (2008). Total and differential bulk cow milk

somatic cell counts and their relation with lipolysis. Livestock Science, 113(2–

3), 274–279. https://doi.org/10.1016/j.livsci.2007.05.007




0302(04)73493-1

Kable, M. E., Srisengfa, Y., Laird, M., Zaragoza, J., Mcleod, J., Heidenreich, J., &

Marco, L. (2016). The core and seasonal microbiota of raw bovine milk in tanker

trucks and the impact of transfer to a milk processing facility. MBio, 7(4).

https://doi.org/10.1128/mBio.00836-16.Editor

Laura, M., Marzotto, M., Dellaglio, F., & Feligini, M. (2009). International journal of

food microbiology study of microbial diversity in raw milk and fresh curd used

for Fontina cheese production by culture-independent methods. International

Journal of Food Microbiology, 130(3), 188–195.


Li, N., Wang, Y., You, C., Ren, J., Chen, W., & Zheng, H. (2018). Variation in raw

milk microbiota throughout 12 months and the impact of weather conditions.

Scientific Reports, 1–10. https://doi.org/10.1038/s41598-018-20862-8

Ma, C., Zhao, J., Xi, X., ng, J. D., Wang, H., Zhang, H., & Kwok, L. Y. (2016).

Bovine mastitis may be associated with the deprivation of gut Lactobacillus.

Beneficial Microbes, 7(1), 95–102. https://doi.org/10.3920/BM2015.0048





1394. https://doi.org/10.1007/s00253-016-7900-2

Patel, R. J., Pandit, R. J., Bhatt, V. D., Kunjadia, P. D., Nauriyal, D. S., Koringa, P.

G., … Kunjadia, A. P. (2017). Metagenomic approach to study the bacterial

89

community in clinical and subclinical mastitis in buffalo. Meta Gene, 12, 4–12.

https://doi.org/10.1016/j.mgene.2016.12.014




https://doi.org/10.3168/jds.2016-11540




https://doi.org/10.1007/s00253-017-8348-8

Wu, H., Nguyen, Q. D., Tran, T. T. M., Tang, M. T., Tsuruta, T., & Nishino, N.

(2019). Rumen fluid , feces , milk , water , feed , airborne dust , and bedding

microbiota in dairy farms managed by automatic milking systems. Anim Sci J.

90(3):445-452. https://doi.org/10.1111/asj.13175





https://doi.org/10.3168/jds.2016-11540

90

CHAPTER 6 Microbiota of manually collected milk between

Chinese, Vietnamese, and Japanese dairy farms

6.1 Introduction

Mastitis is one of the most expensive animal diseases. Generally, the main cause

of mastitis is the invasion of standard pathogens, including Escherichia coli,

Staphylococcus, Streptococcus and so on, into the mammary glands of dairy cows (Bi

et al., 2016). As the main source of bacteria in milk, the bacteria in the environment

will directly affect the bacterial flora in milk. When the feeding mode changes, the

microflora in milk produced by indoor feeding and grazing cows will change

significantly (Doyle et al., 2017). As the main feeding method of modern dairy cows,

the bacteria in milk produced by indoor feeding cows mainly come from dust bacteria

in the air and bacteria on the bedding (Wu et al., 2019). In addition, it is pointed out in

the previous chapters that bacterial flora in milk can change significantly under

different collection methods. Therefore, in order to understand the true flora in the

breast of dairy cows, it is necessary to collect samples manually. However, as the

environment in which mastitis pathogens exist, the theory that only part of dairy cows

living in the same environment will be infected by mastitis pathogens is

incomprehensible.

With the development of science and technology, the methods of microbial flora

analysis in samples are constantly improving. From the past culture-dependent method

to the present culture-independent method, which relies on DNA sequencing analysis,

we have greatly increased the breadth and speed of bacterial flora in samples. High-

throughput sequencing of 16srRNA for bacteria can well describe the composition of

bacterial flora in samples, and further improve our understanding of bacterial flora in

samples. Nowadays, many studies have found that there are also a lot of bacteria in

healthy milk (Bhatt et al., 2012; Bi et al., 2016; Jayarao et al., 2004; Laura et al., 2009;

Li et al., 2018; Zhang et al., 2015). These bacteria in milk will change obviously with

the changes of health level, storage temperature and other conditions of dairy cows

(Kable et al., 2016; Quigley et al., 2013; Rodrigues et al., 2017). However, there is still

a lack of papers directly aimed at the comparison of bacterial flora in healthy milk and

mastitis milk.

On the other hand, the stability of the flora in the mammary glands of dairy cows

91

has not been determined. Milk research is mostly concentrated in a dairy farm, region

or country. Whether the flora in the mammary glands of dairy cows does not have the

difference between countries is still unclear. That is to say, cows of the same breed may

not produce the same quality of milk in different countries. Therefore, in this chapter,

we plan two experiments. The first is to investigate whether the bacterial flora in

mastitis milk is different from that in healthy milk. Experiments 2 investigated whether

milk from cows in different countries had similar bacterial flora.

6.2Materials and methods


6.2.1.1 Collection of Mastitis and Healthy milk

From approximately 40 lactating Holstein cows raised in the Okayama Prefectural

Livestock Research Institute, three cows were randomly selected. All cows were

managed by AMS (Lely Astronout A4, Cornes AG. Ltd., Eniwa, Japan) with exclusive

feeding of total mixed ration silage throughout the year. The contents of dry matter

(DM), crude protein (N × 6.25), and total digestible nutrients in the total mixed ration

silage were 500–600 g/kg, 160–180 g/kg DM, and 720–740 g/kg DM, respectively.

Samples are collected when cows develop mastitis symptoms and healthy milk samples

are collected in the same or similar months. Manual collection was made between

10:00-11:00, and samples from four quarters were mixed to obtain a composite sample.

Following surface cleaning, several streams of foremilk were discarded prior to sample

collection. Automatic collection was made through an AMS device into sterile sample

cups with preservative (bronopol; 2-bromo-2-nitropropane-1,3-diol). Samples from the

latest visit to the AMS were taken as counterparts of the manually collected samples.

The difference in the time of collection between the manual and automatic samplings

was <6 h. All samples were immediately refrigerated on ice and transported to

Okayama University. Milk yield was recorded from the AMS samples, and protein, fat,

solid-not-fat, and the somatic cell count in milk was determined using a CombiFoss

FT+ (Foss Alle, Hillerød, Denmark).

92

6.2.1.2 Collection of milk samples from different countries

From Hei Longjiang in China, Okayama and Hiroshima in Japan and Ho Chi Minh

in Vietnam, a total of six cattle farms were selected for two cattle farms in each country.

Three dairy cows were randomly selected from each dairy farm for analysis. All dairy

cows in China and Japan farms were fed with commercial silage, and cows in Vietnam

farms were fed with formula and fresh grass. In China and Vietnam, milk collection is

managed through the parlor system. In Japan, Automatic milking system (AMS) was

used for milking management. Milk is collected manually and in accordance with the

collection procedure. Before the dairy cows were required to produce milk, the on-duty

veterinarians in the dairy hall used alcohol cotton to clean the nipples and manually

collected and discarded the milk that had just been squeezed under 3-4 times. About 15

ml of milk was collected separately from the breasts of randomly selected five dairy

cows in each farm. The collected milk is kept in ice until it is analyzed.


In experiment 1, nine samples of mastitis milk and nine samples of healthy milk

were used for DNA extraction. And total of 18 samples from each two farms from Hei

Longjiang in China, Okayama and Hiroshima in Japan and Ho Chi Minh in Vietnam

were used for DNA extraction in experiment 2. Bacterial pellets were obtained by







Qiagen, Germantown, MD, USA) according to the manufacturer’s

recommendations(Ni et al., 2017).



93


CMGCCGCGGTAA-3′; reverse:5′-GTGACTGGAGTTCAGACGTGTGCTCTTCCG

ATCTGGACTACHVGGGTWTCTAAT-3′) was employed (Tang, Han, Yu, Tsuruta,


followed by 25 cycles of 94°C for 30 s, 50°C for 30 s, 72°C for 30 s, and a final

elongation of 72°C for 5 min. The products were purified using Fast Gene Gel/PCR



initiation at 94°C for 2 min followed by 10 cycles of 94°C for 30 s, 59°C for 30 s, 72°C

for 30 s and a final elongation of 72°C for 5 min. The PCR products were again purified

as described above. The purified DNA was then ligated to the 16S rRNA amplicons

prior to 250-bp paired-end sequencing performed on an Illumina MiSeq platform at

FASMAC Co., Ltd. (Kanagawa, Japan).










PyNAST.




Tokyo, Japan) to examine the differences between farms and regions. Data for

microbiota and predicted gene functions were used to perform principle coordinate

analysis and heat map construction using Primer version 7 with Permanova+ add-on

software (Primer-E, Plymouth Marine Laboratory, Plymouth, UK). Alpha diversity

indices, i.e. Evenness, observed otus, and Shannon indices, were rarefied to an equal

94

depth of 30,000 sequences by QIIME. LEfSe Analysis of Seasonal Change by

Galaxy(http://huttenhower.sph.harvard.edu/galaxy). The correlation between the milk



6.3 Results

6.3.1 Experiment 1

The reading of bacterial flora in mastitis milk (mean 25898) was significantly

lower than that in healthy milk (mean 39881). At the same time, although the significant

value of alpha diversity analysis is higher than 0.05, it can be clearly seen that the

diversity of bacteria and evenness in healthy milk are higher than those in mastitis milk.

Figure 6.1 Alpha diversity of Bacteria in Milk of Healthy and Mastitis Cows

by one-way ANOVA.

There was no significant difference in phylum level between healthy and diseased

dairy cows. However, in warmer seasons, the percentage of Proteobacteria phylum in

individual milk suddenly increases. In the level of family, the incidence of pathogen-

related family in mastitis milk increased significantly to a very large proportion, such

http://huttenhower.sph.harvard.edu/galaxy

95

as Enterobacteriaceae, Streptococcaceae, Enterococcaceae were found to be a

significant proportion in mastitis milk. In the warm season, especially in the samples

from July to October, the individual differences of bacterial flora in milk were the most

obvious in both healthy cows and mastitis cows. Enterobacteriaceae also appeared in a

large proportion of one healthy milk in July. Meanwhile, pathogenic bacteria, including

Enterobacteriaceae, Streptococcaceae, Enterococcaceae and Staphylococcaceae, were

found in healthy milk.

Similar changes in other bacterial flora occur when a large number of pathogenic

bacteria are present in milk with mastitis. However, in some samples of mastitis, there

is no obvious pathogenic bacteria associated with the bacteriology. On the contrary, the

flora of these mastitis milk was very similar to that of healthy milk samples (>60%). In

the results of one-way ANOVA, the bacterial family in mastitis milk did not have a

higher prevalence than that in a certain bacterial family. However, Staphylococcaceae

content in these samples was generally higher than 1% in these samples. The contents

of Aerococcaceae, Bacteroidaceae and Tissierellaceae in milk of healthy cows were

significantly higher than those of milk with mastitis. Aerococcaceae is nearly five times

more common in healthy milk than in mastitis milk.

96

A

B

Figure 6.2 Phyla level (A) and Family level (B) proportions of Mastitis and healthy

cow’s milk which manually collected in different month managed by an automatic

milking system.

97

A

B C

Figure 6.3 (A) Heatmap showing the relative abundance of major taxa (detected

at >1.0% at least two different samples) (B) Significant changed family in milk

between healthy and mastitis dairy cows by one-way ANOVA. (C) Principal

coordinates analyses of bacterial families in Mastitis and healthy cow’s milk which

manually collected in different month managed by an automatic milking system.

6.3.2 Experiment 2

Two cattle farms in each country were randomly selected and a total of 18 samples

were analyzed. Diversity of the milk microbiota was higher in Japan than in China and

98

Vietnamese cows; the number of operational taxonomy units, Chao 1 index, and

Shannon index were all greater for the milk of Japan than those of China and

Vietnamese cows.

Figure 6.4 Country variation of milk bacterial alpha diversity by one-way ANOVA.

Among the Japanese samples, Bacteroidetes phylum, Cyanobacteria phylum,

Firmicutes phylum, Fusobacteria phylum, Euryarchaeota phylum and Verrucomicrobia

phylum were significantly higher than those of China and Vietnam. Among them,

Sphingobacteriia class in Bacteroidetes phylum and Lactobacillaceae family in

Firmicutes phylum were significantly higher in Chinese samples than in Japanese and

Vietnamese milk samples. Proteobacteria phylum and Actinobacteria phylum were

significantly higher in China and Vietnam than in other countries. At the family level,

the contents of Bacillaceae, Dermacoccaceae and Methylobacteriaceae in Vietnamese

samples were much higher than those in Chinese and Japanese samples. Aerococcaceae

content in Japanese samples was significantly higher than that in other two countries.

Among the Chinese samples, the most distinguishing family is Lactobacillaceae and

Pseudomonadaceae.

99

Figure 6.5 LEfSe analysis at multiple taxonomic levels of each country milk data. Cladogram illustrating the taxonomic groups explaining the most variation among

country. Each ring represents a taxonomic level, with phylum (p), class (c), order (o)

and family (f) emanating from the center to the periphery. Each circle is a taxonomic

unit found in the dataset, with circles or nodes shown in colors (other than yellow)

indicating where a taxon was significantly more abundant. Factorial Kruskal-Wallis

test(p<0.05), pairwise Wilcoxon test(p<0.05) and the LDA log-score(value>2.0) data

was marked on this figure. *: p<0.05; **:P<0.01; ***:P<0.001

100

Relative abundances of Enterobacteriaceae and Propionibacteriaceae were

greater for Chinese and those of Streptococcaceae and Methylobacteriaceae were

greater for Vietnamese milk samples. Japanese milk samples were characterized by low

abundances of potential pathogens. The similarity of bacterial flora in milk samples

from China, Japan and Vietnam is very low (less than 40%). Within the same country,

differences between cattle farms and between cattle and cattle are obvious, but smaller

than differences between countries.


at >1.0% at least two different samples) in milk microbiota from china, Japan and Vietnam.

Clustering was performed using the Euclidean distance as a similarity metric.

101


manually collected milk sampled from china, Japan and Vietnam. The families and

pathways detected at >1% of the total in at least one milk sample are presented.

6.4 Discussion

In the first experiment, all the bacterial family that appeared in large quantities in

mastitis samples were also found in healthy milk. This indicates that under the same

management conditions, all dairy cows have the same probability of invading

pathogenic bacteria. Every cow's milk is attacked by pathogens, but only part of the

cow suffers from disease (Bhatt et al., 2012). This may be due to some external

conditions, such as changes in temperature (season), Day in milk, Metritis, parity, etc.,

may lead to changes in the feeding capacity of dairy cows (Kable et al., 2016; Kayano

et al., 2018; Li et al., 2018; Wittrock et al., 2011). This in turn leads to the production

of milk content changes from the usual ingredients (Kayano et al., 2018). These changes

in milk content can help milk pathogens to proliferate in large quantities while reducing

probiotics that may also inhibit the growth of pathogenic bacteria. As a result, a large

number of pathogenic bacteria related bacteria suddenly appeared in milk. At this time,

102

the number of white blood cells in milk began to increase, some cows will begin to

appear mastitis symptoms. In addition, the tolerance level of dairy cows to pathogenic

bacteria also has individual differences. Some dairy cows, even with a high proportion

of pathogenic bacteria infection, can maintain milk production without mastitis

symptoms (Kayano et al., 2018; Li et al., 2018; Zhang et al., 2015).

In addition, we found that Aerococcaceae, one of the lactic acid bacteria, is

abundant in healthy milk. In previous studies, this bacterium mainly appeared in

Airborne Dust and was the main source of Aerococcaceae in milk (Wu et al., 2019). In

the analysis of the yearly milk flora survey (Chapter 2), the change of the bacteria has

a significant positive correlation with the protein in milk. In Chinese samples (Chapter

5), it was found that there was also a positive correlation between urea nitrogen and

milk. This indicates that the protein content in milk may have a positive correlation

with the breast health of dairy cows. In experiment 2, Aerococcaceae appeared in a

large number of Japanese samples and was significantly higher than that of Chinese and

Vietnamese samples. The study found that the content of Aerococcaceae in milk from

Okayama and Hiroshima was different, and the content of Aerococcaceae in air had

obvious changes. Therefore, we believe that this may be due to the different

management methods which lead to the different content of Aerococcaceae in the

replacement.

Silage is the main feed for dairy cows in China and Japan, while fresh grass is the

main feed for dairy cows in Vietnam. This may be the reason why Vietnamese milk

samples are rich in Actinobacteria phylum. In terms of management, the Chinese and

Vietnamese cattle farms we selected are mainly managed by parlor system. In Japan,

AMS is used for milk management. Parlor system is a milking system requiring manual

participation. Therefore, this may be the reason why milk from China and Vietnam

contains high levels of pathogenic bacteria in milk.

6.5 Conclusion

It is yet unsure if these differences can be due to the differences between AMS and

parlor milking; however, compared with regional difference examined in experiment 4,

meteorological difference could influence on the milk microbiota of dairy cows. Dairy

cow mastitis may not only be caused by pathogenic bacteria infection. The content of

103

milk and the tolerance of cows may have a greater impact on the occurrence of mastitis.

The study in this chapter shows that the diversity of flora in cow breasts under AMS

management will increase significantly. However, the content of pathogenic bacteria in

the breasts of cows under AMS management will be significantly reduced, which is

very powerful in the prevention of mastitis.

6.6 Reference





2672.2012.05244.x

Bi, Y., Wang, Y. J., Qin, Y., Vallverdú, R. G., & García, J. M. (2016). Prevalence of

bovine mastitis pathogens in bulk tank milk in China, PLoS One,

11(5):e0155621. https://doi.org/10.1371/journal.pone.0155621




https://doi.org/10.1128/AEM.02694-16


Guidelines for Monitoring Bulk Tank Milk Somatic Cell and Bacterial Counts.


0302(04)73493-1

Kable, M. E., Srisengfa, Y., Laird, M., Zaragoza, J., Mcleod, J., Heidenreich, J., &

Marco, L. (2016). The core and seasonal microbiota of raw bovine milk in tanker

trucks and the impact of transfer to a milk processing facility. MBio. 7(4).

https://doi.org/10.1128/mBio.00836-16.Editor

Kayano, M., Itoh, M., Kusaba, N., Hayashiguchi, O., Kida, K., Tanaka, Y., … Gröhn,

Y. T. (2018). Associations of the first occurrence of pathogen-specific clinical

mastitis with milk yield and milk composition in dairy cows. Journal of Dairy


104

Laura, M., Marzotto, M., Dellaglio, F., & Feligini, M. (2009). International Journal of

Food Microbiology Study of microbial diversity in raw milk and fresh curd used

for Fontina cheese production by culture-independent methods. International

Journal of Food Microbiology, 130(3), 188–195.


Li, N., Wang, Y., You, C., Ren, J., Chen, W., & Zheng, H. (2018). Variation in raw

milk microbiota throughout 12 months and the impact of weather conditions.

Scientific Reports, 8(1):2371. https://doi.org/10.1038/s41598-018-20862-8





1394. https://doi.org/10.1007/s00253-016-7900-2

Quigley, L., Mccarthy, R., Sullivan, O. O., Beresford, T. P., Fitzgerald, G. F., Ross,



4928–4937. https://doi.org/10.3168/jds.2013-6688




https://doi.org/10.3168/jds.2016-11540




https://doi.org/10.1007/s00253-017-8348-8

Wittrock, J. M., Proudfoot, K. L., Weary, D. M., & von Keyserlingk, M. A. G. (2011).

Short communication: Metritis affects milk production and cull rate of Holstein

multiparous and primiparous dairy cows differently. Journal of Dairy Science,

94(5), 2408–2412. https://doi.org/10.3168/jds.2010-3697

Wu, H., Nguyen, Q. D., Tran, T. T. M., Tang, M. T., Tsuruta, T., & Nishino, N.

(2019). Rumen fluid , feces , milk , water , feed , airborne dust , and bedding

microbiota in dairy farms managed by automatic milking systems, 90(3):445-

452. https://doi.org/10.1111/asj.13175


https://doi.org/10.3168/jds.2016-11540

105




Documents

Investigation of the microbiota of dairy cow milk in ...eprints.lib.okayama-u.ac.jp/files/public/5/56832/...Milk composition and SCC were determined using a CombiFoss FT+, and milk