Johne's Disease Prevention and Control on Organic Dairy Farms in Ontario, Canada

Johne’s Disease Prevention and Control on Organic Dairy Farms in Ontario, Canada

Laura Pieper

A Thesis presented to The University of Guelph

In partial fulfilment of requirements for the degree of Doctor of Philosophy in Population Medicine

Guelph, Ontario, Canada

ABSTRACT

JOHNE’S DISEASE PREVENTION AND CONTROL ON ORGANIC DAIRY FARMS IN ONTARIO, CANADA

Laura Pieper Advisor: University of Guelph, 2014 Professor David F. Kelton

This thesis investigates Johne’s disease (JD) risk factors and control strategies on organic and conventional dairy farms in Ontario, Canada. The JD Risk Assessment and Management Plan

(RAMP) was evaluated and used for the comparison of JD control between both farming types.

Attitudes about JD control among organic producers and veterinarians were further investigated.

RAMP data and JD milk or serum ELISA results from herds voluntarily participating in the

Ontario Johne’s Education and Management Assistance Program (OJEMAP) were used for the first three research chapters. Individual interviews and focus groups with organic producers and veterinarians were used in the last two research chapters to understand attitudes about JD prevention and control, as well as about organic farming and the veterinarian-producer relationship.

The veterinarian conducting the RAMP greatly influenced the RAMP scores and the recommendations that were given to the producers. However, the RAMP was considered useful in determining the between-herd and within-herd JD transmission risk and in identifying recommendations for JD control for the producers.

Organic and conventional farms had a similar herd-level ELISA test-positive prevalence, but affected organic herds had a higher within-herd prevalence than affected conventional herds.

Compared to conventional farms, organic dairy farms were found to have higher risk for JD transmission in the calving and calf management areas, but lower risk in the biosecurity area.

Contrarily, organic producers received fewer recommendations in the calving and calf management areas.

There was hesitation among organic producers to change management practices for JD control and among veterinarians to recommend certain management changes because of organic practices. Organic producers tended to focus on test-and-cull strategies, whereas veterinarians focused on management improvement to control the spread of the disease. While the veterinarian-organic farmer relationships were mostly good, most veterinarians were lacking knowledge about organic farming and did not appear to be the main advisor on many operations.

Therefore, education efforts for organic dairy producers regarding approaches for JD control should be increased. Continuing education for veterinarians regarding organic dairy production and regulations might help in improving the veterinary-client relationship and, consequently, in delivery of animal health programs on organic dairy farms.

Meiner Familie

ACKNOWLEDGEMENTS

I would like to thank all the people who helped me accomplish this research. Writing this section has been the most difficult part of the entire thesis. It seems impossible to put into words what each one of you means to me and how you helped me during the process.

First and foremost, I would like to thank my supervisor Dr. Dave Kelton for the continuous support in all matters of this research and graduate experience. I am grateful for the trust and freedom that I received from you. During my time in Guelph, I had the opportunity to learn, explore, travel and get to know the scientific community. My experiences were beyond any expectations.

I also want to thank my PhD committee, including Drs. Ann Godkin, Ulrike Sorge (Riki), Trevor

DeVries, and Kerry Lissemore for their input, advice and support throughout this research.

Thank you to all organic dairy producers and veterinarians who participated in the interviews and focus groups. You spent your precious time honestly answering my questions and I hope the results of this research will ultimately help you and others in the industry in controlling and preventing Johne’s disease. Thank you also to Jenny Butcher and Shelly Juurlink from Organic

Meadow for their valuable input throughout this project.

This project could not have been completed without the financial support through the Ontario

Ministry of Food (OMAF) – University of Guelph Research Partnership. The project was additionally supported by an NSERC travel scholarship. Supplementary personal financial support was kindly provided through an Ontario Graduate Scholarship, University International

Graduate Scholarships and private donors. I would like to thank all donors for their generosity.

Thank you, Prof. Staufenbiel, for your suggestion and encouragement to come to Guelph to pursue a PhD in Epidemiology. It turned out as a great experience.

Many thanks go to Laura Falzon, Andreia Arruda, Terri Ollivett, Bimal Chhetri, and Khaled

Gohary for being such wonderful friends. Thank you also to my other friends and colleagues at

Population Medicine, including Jessica and Troy Gordon, Raphael Neves, Cynthia Miltenburg,

Christine Murray, Shannon Meadows. I enjoyed the time I was able to spend with you and will work hard to keep in touch. The long discussions about life, culture, and pretty much anything else gave me the opportunity to learn and appreciate different views on the world around me. I believe I grew as a person because of each and every one of you.

Thank you to Jill and Jairo Melo and the Mini-Melos for providing a place to live and a home away from home. I will never forget the late-night pool parties and evenings we celebrated together.

My special thanks go to Antonio who put up with me for the last 4 years. Thank you for being there to inspire me, to celebrate with me and to lift my spirit when I was down.

All of this would have not been possible without the support and encouragement from my parents, Bernd and Marlies, my siblings, Robert and Jule, and my grandparents, Hanni, Ulli and

Betty. Thank you for being there for me at all times.

vii

STATEMENT OF WORK DONE

The Ontario Johne’s Education and Management Assistance Program (OJEMAP) was a program conducted independently from this thesis. Nicole Perkins entered most of the RAMP data into a database and Jamie Imada coded the verbal RAMP recommendations into numerical codes. Ann

Godkin conducted and coordinated most of the education efforts for producers and veterinarians.

Milk and serum samples were analysed for JD ELISA test-positivity by CanWest Dairy Herd

Improvement (DHI) and Animal Health Laboratory (University of Guelph, Canada), respectively.

Program documents provided in Appendix I.I and I.II were produced by the OJEMAP working group; however, they were included in this thesis to ensure that the context of this work is understandable, even if the documents may be changed or unavailable in the future.

The initial proposal for this study was prepared by David Kelton, Ulrike Sorge, Ann Godkin,

Gaston Raggio, and Shelly Juurlink. The human ethics approval was prepared by Laura Pieper with the help of David Kelton, Ann Godkin, Trevor DeVries, and Kerry Lissemore.

Data management and all statistical analysis were conducted by Laura Pieper. Karen Hand gathered further herd data from CanWest DHI and Dairy farmers of Ontario (DFO). Geographic display of herd participation in CHAPTER 2 was also prepared by Karen Hand. Laura Pieper contacted potential participants for CHAPTERS 5 and 6, conducted all individual interviews and transcribed audio-recordings from interviews and focus groups. The focus groups were conducted by Laura Pieper and Daniel Shock. Qualitative analysis in CHAPTERS 5 and 6 were conducted by Laura Pieper with help from Andria Jones-Bitton.

All chapters were written entirely by Laura Pieper and reviewed by all committee members including David Kelton, Ann Godkin, Ulrike Sorge, Trevor DeVries, and Kerry Lissemore.

viii

TABLE OF CONTENTS

CHAPTER 1 MANAGEMENT PRACTICES AND THEIR POTENTIAL INFLUENCE ON JOHNE’S DISEASE TRANSMISSION ON CANADIAN ORGANIC DAIRY FARMS – A CONCEPTUAL ANALYSIS ...... 1 1.1 Abstract ...... 1 1.2 Introduction ...... 2 1.3 Canadian organic dairy farming ...... 3 1.3.1 Canadian Organic Standards ...... 3 1.3.2 Organic dairy industry ...... 5 1.3.3 Scientific investigations focused on organic dairy production ...... 6 1.4 Johne’s disease ...... 9 1.4.1 Etiology and epidemiology ...... 9 1.4.2 Diagnostic tests ...... 11 1.4.3 Johne’s disease prevalence and control in Canada ...... 14 1.5 Johne’s disease prevalence on organic dairy farms...... 16 1.6 Association between Johne’s disease and organic dairy farming ...... 17 1.6.1 Farm structure ...... 17 1.6.2 Crop management ...... 20 1.6.3 Nutrition ...... 20 1.6.4 Calving and dairy heifer management ...... 23 1.6.5 Veterinary treatments ...... 26 1.6.6 Biosecurity ...... 28 1.6.7 Breeding strategy ...... 29 1.7 Knowledge gaps ...... 31 1.8 Conclusions ...... 33 1.9 Acknowledgments ...... 33 1.10 Objectives ...... 33 1.11 References ...... 34 CHAPTER 2

VARIABILITY IN THE RISK ASSESSMENT AND MANAGEMENT PLAN (RAMP) SCORES COMPLETED AS PART OF THE ONTARIO JOHNE’S EDUCATION AND MANAGEMENT ASSISTANCE PROGRAM (2010-2013) ...... 49 2.1 Abstract ...... 50 2.2 Introduction ...... 50 2.3 Materials and Methods ...... 53 2.3.1 Data ...... 53 2.3.2 Statistical analyses ...... 53 2.4 Results ...... 54 2.5 Discussion ...... 56 2.6 Conclusion ...... 58 2.7 References ...... 59 2.8 Tables ...... 60 2.9 Figures ...... 63 CHAPTER 3 EVALUATION OF THE JOHNE’S DISEASE RISK ASSESSMENT AND MANAGEMENT PLAN IN DAIRY HERDS IN ONTARIO, CANADA ...... 65 3.1 Abstract ...... 65 3.2 Introduction ...... 66 3.3 Materials and Methods ...... 69 3.4 Results ...... 72 3.5 Discussion ...... 79 3.6 Conclusion ...... 84 3.7 Acknowledgements ...... 84 3.8 References ...... 85 CHAPTER 4 COMPARING ELISA PREVALENCE, RISK FACTORS AND MANAGEMENT RECOMMENDATIONS FOR JOHNE’S DISEASE PREVENTION BETWEEN ORGANIC AND CONVENTIONAL DAIRY FARMS IN ONTARIO ...... 89 4.1 Abstract ...... 89 4.2 Introduction ...... 90 4.3 Materials and Methods ...... 94

4.3.1 Participation of organic herds ...... 94 4.3.2 Data ...... 94 4.3.3 Statistical analysis ...... 94 4.4 Results ...... 97 4.5 Discussion ...... 100 4.6 Conclusion ...... 105 4.7 Acknowledgements ...... 105 4.8 References ...... 106 4.9 Tables ...... 110 CHAPTER 5 UNDERSTANDING JOHNE’S DISEASE CONTROL AND DEVELOPMENT OF RECOMMENDATIONS FOR JOHNE’S DISEASE PREVENTION ON ONTARIO ORGANIC DAIRY FARMS ...... 117 5.1 Abstract ...... 117 5.2 Introduction ...... 118 5.3 Materials and Methods ...... 120 5.3.1 Questionnaire ...... 120 5.3.2 Interviews ...... 120 5.3.2.1 Interview design ...... 120 5.3.2.2 Interview participants ...... 121 5.3.2.3 Interview structure ...... 121 5.3.2.4 Data analysis ...... 122 5.4 Results ...... 122 5.4.1 Questionnaire ...... 122 5.4.2 Interviews ...... 123 5.4.2.1 Overview ...... 123 5.4.2.2 Importance of organic farming ideals ...... 124 5.4.2.2.1 Natural and conventional livestock management ...... 124 5.4.2.2.2 Biosecurity ...... 128 5.4.2.2.3 Hesitation to change due to being an organic producer ...... 130 5.4.2.2.4 Veterinarian’s understanding of organic farming ...... 131 5.4.2.3 Importance of JD ...... 132

5.4.2.3.1 Association between JD and Crohn’s disease ...... 132 5.4.2.3.2 Consumer perception and risks ...... 132 5.4.2.3.3 Negative effects of JD on dairy farming ...... 134 5.4.2.4 Understanding of the JD control program ...... 134 5.4.2.4.1 Attitudes about the RAMP ...... 134 5.4.2.4.2 Comparison with other disease control programs ...... 135 5.4.2.4.3 Importance of testing and JD prevalence ...... 136 5.4.2.4.4 Prevention strategy depends on prevalence ...... 137 5.4.2.4.5 Information and knowledge ...... 138 5.4.2.5 Suggestions and recommendations ...... 139 5.4.2.5.1 Organic farming ideals ...... 139 5.4.2.5.2 Importance of JD ...... 140 5.4.2.5.3 Understanding of the JD control program ...... 140 5.5 Discussion ...... 140 5.6 Conclusion ...... 145 5.7 Acknowledgements ...... 146 5.8 References ...... 146 5.9 Tables ...... 148 5.10 Figures ...... 152 CHAPTER 6 UNDERSTANDING ONTARIO ORGANIC DAIRY FARMERS AND THE RELATIONSHIP WITH THEIR HERD VETERINARIAN ...... 154 6.1 Abstract ...... 154 6.2 Introduction ...... 155 6.3 Materials and Methods ...... 157 6.3.1 Study design ...... 157 6.3.2 Study participants ...... 157 6.3.3 Interview structure ...... 158 6.3.4 Data analysis ...... 158 6.3.5 Researcher’s perspective ...... 159 6.4 Results ...... 159 6.4.1 Participant’s demographics ...... 159

xii

6.4.2 Organic farming ...... 160 6.4.2.1 Organic farming overview ...... 160 6.4.2.2 Holistic farming ...... 161 6.4.2.3 People ...... 162 6.4.2.3.1 Consumers ...... 162 6.4.2.3.2 Family ...... 162 6.4.2.3.3 Self ...... 163 6.4.2.3.4 Social networks ...... 164 6.4.2.4 Administrative ...... 166 6.4.2.4.1 Financial ...... 166 6.4.2.4.2 Paperwork ...... 167 6.4.2.4.3 Marketing ...... 167 6.4.2.5 Environment ...... 168 6.4.2.5.1 Environment ...... 168 6.4.2.5.2 Soil and crops ...... 169 6.4.2.6 Animals ...... 170 6.4.2.6.1 Animal health ...... 170 6.4.2.6.2 Animal welfare ...... 174 6.4.3 Veterinarian-Producer relationship ...... 176 6.4.3.1 Veterinarian-Producer relationship overview ...... 176 6.4.3.2 Personal ...... 176 6.4.3.2.1 Veterinarian-Producer personal relationship ...... 176 6.4.3.2.2 Veterinarian’s attitudes about organic farming ...... 177 6.4.3.2.3 Veterinarian’s attitudes about alternative medicine ...... 179 6.4.3.3 Veterinary profession ...... 181 6.4.3.3.1 Veterinarian-Producer working relationship ...... 181 6.4.3.3.2 Veterinarian ideal ...... 184 6.4.4 Knowledge and information seeking behavior ...... 186 6.4.4.1 Knowledge and information seeking behavior overview ...... 186 6.4.4.2 Knowledge about organic dairy production ...... 186 6.4.4.3 Information seeking behavior of producers ...... 186 6.4.4.4 Information seeking behavior of veterinarians ...... 188

xiii

6.5 Discussion ...... 188 6.6 Conclusion ...... 192 6.7 Acknowledgements ...... 192 6.8 References ...... 192 6.9 Tables ...... 195 6.10 Figures ...... 197 CHAPTER 7 DISCUSSION ...... 199 7.1 General discussion and limitations ...... 199 7.2 Conclusions ...... 206 7.3 Future directions ...... 207 7.4 References ...... 209 APPENDIX ...... 212 APPENDIX I.I ...... 212 APPENDIX I.II ...... 214 APPENDIX II ...... 222 APPENDIX III.I ...... 227 APPENDIX III.II ...... 227 APPENDIX III.III ...... 227 APPENDIX III.IV ...... 231 APPENDIX IV.I ...... 234 APPENDIX IV.II ...... 234 APPENDIX IV.III ...... 237 APPENDIX V.I ...... 242 APPENDIX V.II ...... 245

xiv

LIST OF TABLES

Table 1.1 Sensitivity and specificity of diagnostic tests for JD ...... 13

Table 1.2 Description of Johne's disease control programs in Canadian provinces and regions (Ron Barker, unpublished) ...... 14

Table 1.3 Johne's disease dairy herd-level prevalence in Canadian provinces ...... 15

Table 2.1 Distribution of Risk Assessment and Management Plans (RAMPs) at the county, veterinary clinic and veterinarian level ...... 60

Table 2.3 Most frequent recommendations for Johne’s disease prevention and control given by veterinary assessors to producers after the Risk Assessment and Management Plan (RAMP) ...... 62

Table 2.4 Final mixed linear regression model predicting the overall RAMP score in the herd. Herds cluster by assessing veterinarian (Intra-class Correlation Coefficient=0.24) ...... 62

Table 3.3 Zero-inflated negative binomial models A and B predicting the number of ELISA positive animals in an Ontario dairy herd...... 78

Table 4.1 Description of herd characteristics and milk quality parameters in conventional and organic Ontario dairy herds ...... 110

Table 5.1 Number of responses (proportion) for questionnaire about experience with Risk Assessment and Management Plan (RAMP) on Ontario organic dairy farms (n=20) ... 148

Table 5.2 Number of participants in individual interviews and focus groups about Johne’s disease prevention on organic dairy farms ...... 148

xvi

Table 6.1 Regulations about medical treatments of certified organic animals as described in the Canadian Organic Standards (COS; Government of Canada, 2006) ...... 195

Table 6.2 Demographic information of interviewees ...... 196

Table 6.3 Organic producer's and veterinarian's accuracy of knowledge about organic farming regulations (SD=standard deviation; Min=minimum; Max=maximum) ...... 197

xvii

LIST OF FIGURES

Figure 1.1 Canadian organic label for products with >95% certified organic content and that has been certified under Canadian organic requirements ...... 3

Figure 1.2 Growth of organic dairy industry between 2000 and 2010. (Data from Agriculture and Agri-Food Canada, 2009) ...... 6

Figure 1.3 Average herd size in organic and conventional Canadian dairy herds (data from Stonehouse et al., 2001, Rozzi et al., 2007, Roberts et al., 2008, Pieper et al., 2013) ..... 18

Figure 2.1 Flowchart of number of observations at each stage from combining the RAMP data with the ELISA data until the final number of observations analyzed ...... 63

Figure 2.2 Choropleth maps of number of herds per county and proportion of herds per county participating in the Ontario Johne’s Education and Management Assistance Program in southern Ontario ...... 64

Figure 3.1 Predicted probability of receiving a recommendation for a section by respective RAMP section score and Herd ELISA Status (HES); panel a=recommendation for general Johne’s disease risk management, panel b=recommendation for calving area risk management, panel c=recommendation for pre-weaned heifer risk management, panel d=recommendation for weaned heifer to first calving risk management, panel e=recommendation for adult cow risk management; x-scales were adjusted according to the range of the section scores ...... 76

Figure 3.2 Comparison of two zero-inflated negative binomial models (Model A and Model B); Upper Panel) Observed and predicted probability of count of ELISA positive cows per herd; Lower Panel) Difference between observed and predicted probability of count of ELISA positive animals per farm for Model A and Model B ...... 79

Figure 5.1 Thematic map of Johne's disease prevention and control on organic dairy farms in Ontario, Canada. Bold font= themes; normal font= subthemes ...... 152

Figure 5.2 Adoption of management practices for JD prevention explained by the Health Belief Model. Panel A) Low perceived susceptibility (e.g. test negative herd status) and

xviii

consequently low likelihood of management change for JD prevention; Panel B) High perceived susceptibility (e.g. test positive herd status) and consequently high likelihood of management change for JD prevention. Bold lines indicate a strong connection, straight lines a medium connection, dashed lines a weak connection, arrows pointing up indicate high and arrows pointing down indicate low...... 153

Figure 6.1 Thematic map of producers’ attitudes and challenges regarding organic dairy farming ...... 197

Figure 6.2 Thematic map about veterinarians’ and organic dairy producers’ perceptions of their relationship ...... 198

Figure 6.3 Thematic map of knowledge and information seeking behavior of veterinarians and organic dairy producers ...... 198

xix

LIST OF ABBREVIATIONS

Abbreviation Explanation AIC Akaike Information Criteria BIC Bayesian Information Criteria BRD Bovine Respiratory Disease BSE Bovine Spongiform Encephalophathy BtSCC Average Bulk Tank Somatic Cell Count BVD Bovine Viral Diarrhea CAD Canadian Dollar CFIA Canadian Food Inspection Agency CFU Colony Forming Units CI Confidence Interval CJDI Canadian Johne’s Disease Initiative COS Canadian Organic Standards CR Count Ratio DFO Dairy Farmers of Ontario DHI Dairy Herd Improvement DNA Deoxyribonucleic Acid ELISA Enzyme-linked immunosorbent assay FMD Foot and Mouth Disease GMO Genetically Modified Organism HES Herd ELISA Status HP High Positive animal IBR Infectious Bovine Rhinotracheitis ICC Intra-class Correlation Coefficient IQR Inter Quartile Range JD Johne’s disease Lntotnotest Log transformed number of animals tested MAP Mycobacterium avium spp. paratuberculosis n Number of observations NAHMS National Animal Health Monitoring System OABP Ontario Association of Bovine Practicioners OD Optical Density (milk ELISA) OJEMAP Ontario Johne’s Education and Management Assistance Program OR Odds Ratio p Significance level PCR Polymerase Chain Reaction QN Question QREC Recommendation for a RAMP question RAMP Risk Assessment and Management Plan S:P Sample to Positive Ratio (serum ELISA) SCC Somatic Cell Count SD Standard Deviation SE Standard Error of the mean SEC Section SREC Recommendation for a RAMP section ZINB Zero-Inflated Negative Binomial (regression model)

CHAPTER 1

MANAGEMENT PRACTICES AND THEIR POTENTIAL INFLUENCE

ON JOHNE’S DISEASE TRANSMISSION ON CANADIAN ORGANIC

DAIRY FARMS – A CONCEPTUAL ANALYSIS

Laura Pieper1, Ulrike S. Sorge2, Ann Godkin3, Trevor DeVries4, Kerry Lissemore1, David

Kelton1

1Department of Population Medicine, University of Guelph, Ontario, Canada, N1G 2W1

2Department of Veterinary Population Medicine, University of Minnesota, St. Paul, USA, MN

55108

3Veterinary Science and Policy Group, Ontario Ministry of Agriculture and Food (OMAF),

Ontario, Canada, NOB 1S0

4Department of Animal and Poultry Science, University of Guelph, Ontario, Canada, N1G 2W1

1.1 ABSTRACT

Johne’s disease (JD) is a chronic, production-limiting disease of ruminants. Control programs aiming to minimize the effects of the disease on the dairy industry have been launched in many countries, including Canada. Those programs commonly focus on strict hygiene and management improvement, often combined with various testing methods. Concurrently, organic dairy farming has been increasing in popularity. Because organic farming promotes traditional management practices, it has been proposed that organic dairy production regulations might interfere with implementation of JD control strategies. However, it is currently unclear how organic farming would change the risk for JD control. This review presents a brief introduction to organic dairy farming in Canada, JD, and the Canadian JD control programs. Subsequently, organic practices are described and hypotheses of their effects on JD transmission are developed. Empirical

research is needed, not only to provide scientific evidence for organic producers, but also for smaller conventional farms employing organic-like management practices.

1.2 INTRODUCTION

The organic dairy sector is a rapidly growing industry in Canada (Agriculture and Agri-Food

Canada, 2012). To receive organic certification, farmers must adjust their management practices to comply with the national Canadian Organic Standards (COS, Government of Canada, 2006).

Additionally, some organic farmers further employ altered management practices reflecting their beliefs. Due to restrictions that the standards impose on treatment options for diseased animals, organic farms need to place more emphasis on prevention rather than treatment of diseases

(Government of Canada, 2006). However, the unique circumstances on organic farms have the potential to modify infectious disease transmission risks among and within herds.

The aim of this review is to raise awareness about the risks for Johne’s disease (JD) transmission and opportunities for its prevention on organic dairy farms in Canada, with the objective of identifying areas that might need specific research and management attention. This review focuses on organic dairy cattle farming; organic dairy sheep, dairy goat, and beef cattle farming were omitted. An organic dairy farm was defined as a farm certified under organic regulations in the respective country. Parts of the COS and organic practices were compared with current recommendations for JD prevention and control. Comments were made about how, in the authors’ opinion, the risk for disease transmission could be impacted under Canadian organic dairy farming conditions.

1.3 CANADIAN ORGANIC DAIRY FARMING

1.3.1 Canadian Organic Standards Organic farming evolved as a counter movement to the increased use of technology in farming, at the beginning of the 20th century (Hill and MacRae, 1992). Organic production is now formally legislated in Canada by the regulations CAN/CGSB-32.310-2006 Organic Production Systems -

General Principles and Management Standards (COS, Government of Canada, 2006), the

CAN/CGSB-32.311-2006 Organic Production Systems - Permitted Substances Lists

(Government of Canada, 2006), and the Organic Products Regulations, 2009 (SOR/2009-176)

(Government of Canada, 2009). To sell organic milk inter-provincially, internationally, or to use the organic logo (Figure 1.1) farmers must be certified under these regulations (Government of

Canada, 2009). The specifically stated aim of organic production is to optimize productivity in diverse situations and design a sustainable system within the agro-ecosystem. It focuses on protecting the environment, minimizing soil degradation, maintaining biological diversity, recycling resources within the farm, and caring for animals to promote their health and to meet their behavioral needs (Government of Canada, 2006).

Figure 1.1 Canadian organic label for products with >95% certified organic content and that has been certified under Canadian organic requirements

Currently, Canadian Food Inspection Agency (CFIA) designated Conformity Verification Bodies accredit third party certifiers to ensure the uniform application of the COS (Government of

Canada, 2009). However, since the COS were only implemented in 2006 and the Organic

Products Regulation in June 2009, many producers, veterinarians and advisors are not yet familiar with the specifics. For example, it is widely believed among dairy professionals that antibiotic therapy is entirely forbidden for organic dairy production. However, the COS state that necessary treatment should not be withheld from sick animals. In emergency situations when alternative treatments (herbal or homeopathic) are ineffective, antibiotics can be used for up to two cases per lactation and the treated animal can return into organic production after 30 days or twice the legal withdrawal time, whichever is longer. Prophylactic or metaphylactic antibiotics are not allowed (Government of Canada, 2006). Similar regulations are in place for hormone or parasiticide treatments. Furthermore, veterinarians are supposed to play a key role in disease prevention and control, and written instructions for the chemical allopathic treatments have to be given to the producer (Government of Canada, 2006). Because COS put restrictions on the treatment options for sick animals, farmers are required to focus more on prevention rather than treatment of diseases (Government of Canada, 2006).

Certified organic production animals have to be fed organic feed. The COS limit the amount of grain fed to herbivores and specify that the forage intake from grazing has to be at least 30% of the dry matter consumed during the pasture season. In the winter, cows must have regular exercise preferably daily or at the very least twice per week.

Housing and nutrition for dairy calves is also regulated. Organically raised calves have to receive natural cow milk until weaning; however, suckling from the dam is not a requirement in the COS.

Theoretically, organic milk replacer could be fed in cases of emergency, but it is currently unavailable. There is also no specification that calves have to stay with the dam after calving; but, group housing with other calves after the first three months of life is required. Similar to adult dairy cows, dairy heifers, older than nine months of age, have to be turned out on pasture, depending on the season (Government of Canada, 2006).

1.3.2 Organic dairy industry The organic dairy industry in Canada has been growing steadily for more than a decade. Between

2000 and 2010, organic milk production has increased by almost 10-fold. At the same time, the number of producers has increased gradually (Figure 1.2). In 2011/2012, 1.2% of the total national milk output was produced by 218 certified organic farmers. The province with the greatest production of organic milk and the highest number of organic dairy farmers is Quebec, followed by Ontario, British Columbia, and Alberta. British Columbia has the largest proportion of milk produced organically (2.8%), followed by Alberta (1.4%), Quebec (1.3%), and Ontario

(1.0%) (Agriculture and Agri-Food Canada, 2012).

This production does not necessarily reflect the amount of milk processed to produce organic foods and beverages. Due to underdeveloped infrastructure in parts of the country, some organic milk still goes into the conventional milk processing stream. In Ontario, about 10% of organically produced milk is processed as conventional milk and producers do not receive the incentive for organic milk, even though the organic milk market is relatively well established. Furthermore, small amounts of organic milk are imported, mainly from the USA (Agriculture and Agri-Food

Canada, 2012).

Figure 1.2 Growth of organic dairy industry between 2000 and 2010. (Data from Agriculture and Agri-Food Canada, 2009)

Although producers are rewarded for their efforts to produce organic milk with premium payments for milk classes 1 and 2 of $0.25 to $0.30 CAD per litre (depending on the province), they often need to pay $0.08 to $0.12 CAD per litre in additional administration and transportation fees that are commonly deducted from this premium prior to payment (Agriculture and Agri-Food Canada, 2012).

1.3.3 Scientific investigations focused on organic dairy production Scientific literature targeting organic dairy farming in Canada is sparse. The majority of published articles focus on economic or environmental impacts of organic dairy farming. One article (Rozzi et al., 2007) investigates breeding values for organic dairy cattle. To date, little has been published about animal health or welfare on organic dairy farms in Canada.

In Ontario, organic farms usually have lower milk production per cow and per hectare than conventional herds (Roberts et al., 2008). These findings are supported by the earlier Ontario study from Stonehouse et al. (2001) but not by Ogini et al. (1999). Ogini et al. (1999) and

Stonehouse et al. (2001) showed that although organic farmers have fewer crop sales revenues, or

fewer crop and milk sales revenues per hectare, respectively, the majority of them are better off economically because they tend to rely more heavily on self-sufficiency rather than on off-farm inputs. This self-sufficiency is represented, for example, in raising their own replacement stock, growing their own livestock feed, as well as no use of chemical fertilizer, herbicides, or pesticides.

In multiple studies, it was reported that Ontario organic farms have a high proportion (>50%) of their land base utilized for forage production for livestock (Roberts et al., 2008; Sholubi et al.,

1997, Rozzi et al. 2007). Farm nutrient (N, P, K) surpluses per hectare per year were lower than reported on conventional farms, and soil P levels were low on approximately half of the farms whereas exchangeable K levels were moderate to high on all farms. Some organic farms showed a negative P balance, which indicates that depletion of soils could happen if insufficient nutrient supplementation is done (Roberts et al., 2008).

To optimize nutrient input and utilization of the soil, organic farms employ long rotation cycles including pasture, legumes and small grains, instead of the simple corn-soy rotations which are used on many conventional farms. They also compost the liquid and solid manure before applying it to the crops (Sholubi et al., 1997). Organic farmers tend to preserve, or even build up, soil organic matter to increase soil fertility, and retain water (Lynch, 2009).

Sholubi et al. (1997) reported that organic farmers did not like the application or the high cost of agrochemicals, and were concerned about the environmental pollution associated with those artificial inputs. In 2010, Cranfield et al. reported that economic, environmental, health and safety, ideological and philosophical, as well as other motivations play a role in converting to organic agriculture (dairy and vegetable). The aforementioned authors also describe challenges during and after transition related to marketing, economic aspects and infrastructure. Personal, environmental, economic, health and safety, and animal welfare benefits are also identified.

Although it was shown that organic farms tend to be more profitable, economic benefits were not the most important reasons for converting to organic farming. The most important reasons included environmental, animal and human health and safety aspects of their practices over the long term (Cranfield et al. 2010; Sholubi et al., 1997).

Milk yield on organic farms is usually lower and the herd size is comparable or smaller than on conventional farms (Roberts et al., 2008, Stonehouse et al., 2001, Rozzi et al., 2007). An indicator for milk quality is the number of white blood cells in the milk, commonly referred to as somatic cell count or SCC, with low SCC indicating good milk quality. Rozzi et al. (2007) reported the average SCC of organic herds was 309,000 cells/mL, or about 50,000 cells/mL higher than on conventional farms, and there was an inverse relationship between milk yield and

SCC. Therefore, organic herds with intensive management and high milk yield were found to have lower SCC than organic herds with less intensive management and lower milk yield. In a survey of 8 farmers in western Ontario, the most commonly farmer-reported livestock diseases were mastitis, milk fever, and calf pneumonia. Interestingly, all eight of those farmers used homeopathy either alone or in combination with probiotics or chemical allopathic veterinary drugs to treat these diseases (Sholubi et al., 1997). In an economic analysis of those farms by

Stonehouse et al. (2001) organic producers reported lower expenses for veterinary costs, drugs or breeding compared to their conventional counterparts.

The majority of Ontario organic producers keep the Holstein breed; however, genetic selection for traits such as longevity, health, and grazing capability are emphasized more than production.

Many farmers keep other purebred breeds (Brown Swiss or Jersey) or experiment with crossbreeding Holstein with Dutch Belted, Brown Swiss, Jersey, or Milking Shorthorn breeds in an effort to increase the animals’ robustness and fitness. In organic herds, cross breeds and the use of a breeding bull are more common in herds with lower milk production than herds with higher

milk production. The use of a breeding bull is also more common if minor or rare breeds are used because semen for artificial insemination is generally unavailable (Rozzi et al., 2007).

1.4 JOHNE’S DISEASE

1.4.1 Etiology and epidemiology Johne’s disease (JD) is a chronic, infectious gastrointestinal disease caused by Mycobacterium avium spp. paratuberculosis (MAP). MAP has been shown to infect a wide variety of domestic and wild ruminants such as cattle, sheep, goats, deer (Machackova et al., 2004, Nebbia et al.,

2000) and mouflons (Machackova et al., 2004). However, they have also been isolated from monogastric mammals such as rabbits (Machackova et al., 2004; Nugent et al., 2011, Judge et al,

2005) and hedgehogs (Nugent et al., 2011), as well as from wild birds residing close to or on farm property (Nugent et al., 2011).

MAP belong to the genus Mycobacterium and are, therefore, related to Mycobacterium tuberculosis and Mycobacterium bovis, which cause tuberculosis in humans and cattle, respectively. Significantly, MAP shares microbiological and pathogenic properties with those bacterial species. It is a very slow growing, acid-fast bacillus. Contrary to the other aforementioned species and its closer relative Mycobacterium avium spp. avium, MAP must be supplemented with the iron transport chemical mycobactin to grow. For this reason, it is an obligate intracellular organism (Rolle and Mayr, 2007).

As summarized by Tiwari et al. (2006), disease progression can be divided into four phases. In the first silent phase, in animals less than 2 years of age, detection of infection with cost effective measures is not possible and clinical or subclinical effects are absent. In the subsequent second phase, or subclinical period, the infection can be detected by diagnostic tests (directly by fecal culture or PCR, or indirectly by serum or milk antibody) and can be associated with a drop in milk yield. In the third phase or clinical period, the animals show typical signs indicative of MAP 9

infection. The clinical signs in cattle include chronic or intermittent diarrhea, gradual weight loss despite normal appetite, and loss in milk production. The infection can be detected by fecal culture/PCR or antibody ELISA. In the advanced clinical infection stage, affected animals, if not culled for this or other reasons, will develop hypoproteinemia, submandibular edema (bottle jaw) due to protein loss, and emaciation. Unless euthanized, animals reaching this phase eventually die due to cachexia.

Infected animals shed the bacteria mainly in the feces; however shedding via milk (Streeter et al.,

1995), semen (Khol et al., 2010; Buergelt et al., 2004; Ayele et al. 2004; Larsen et al. 1981,

Larsen and Kopecky, 1970), and saliva (Sorge et al., 2013) have also been reported. Likely, the most common route of disease transmission is by ingestion of very small amounts of contaminated feces or milk early in a calf’s life. The risk of a MAP-infected calf developing JD is particularly high for calves exposed at calving, where the calving area is highly contaminated

(Windsor and Whittington, 2010). For experimental infection, intravenous and subcutaneous inoculation has been used successfully (Mitchell et al., 2011). It is unclear whether it is possible to infect cows with contaminated semen. The studies of Merkal et al. (1982) and Owen and

Thoen (1983) showed positive antibody reaction, abortions, and likely antibody or MAP-positive calves when inoculating large concentrations of MAP into the uterus after artificial insemination.

However, the test concentrations (5x108cfu MAP in 5ml) used by the authors are much higher than the concentrations usually found in semen (102-105/ml, Khol et al., 2010). Recently,

Whittington and Windsor (2009) published a systematic review of intrauterine transmission from infected dams to the unborn calves. According to the authors, 39% (20-60%) and 9 % (6-14%) of calves born to dams with clinical and subclinical JD, respectively, were JD-positive themselves.

Susceptibility to infection is highest when the animals are young and decreases as they age

(Windsor and Whittington, 2010). Therefore, most of the JD prevention efforts focus on calving

and early calf rearing practices. Having cows calve in individual calving pens rather than group calving pens is associated with decreased odds of being a JD positive farm (Wells and Wagner,

2000; Tiwari et al., 2009). It is commonly recommended to remove the calf from the calving area as soon as possible (<12h) after birth (Wraight et al., 2000). Other authors report access of calves less than 6 months of age to adult cows as being a major risk factor for JD transmission (Marcé et al., 2011; Diéguez et al., 2008). As summarized by Mitchell et al. (2011) calves infected at <3 months of age were able to shed MAP beginning one month after the exposure, posing the risk of calf-to-calf transmission in group housing situations (van Roermund et al., 2007). Furthermore, the source of colostrum and milk has been associated with JD transmission. Calves that receive colostrum from multiple cows rather than only from their own dam are at increased risk (odds ratio of 1.24) of being JD positive (Nielsen et al., 2008), and calves fed colostrum replacer are less likely to become infected with MAP compared to calves fed maternal colostrum at birth

(Pithua et al., 2009). Similar associations hold true for the milk feeding period. Calves suckling a foster cow compared to receiving artificial milk replacer have increased odds of being JD positive (Nielsen et al., 2008).

It has been demonstrated that the odds of being JD-positive is greater for larger herds than for smaller herds (Muskens et al., 2003; Wells and Wagner, 2000). Furthermore, farms that purchase stock from other farms are more likely to be positive than farms that do not purchase stock

(closed herds) (Wells and Wagner, 2000; Tiwari et al., 2009).

1.4.2 Diagnostic tests Typically, infected animals shed the bacteria before they show clinical signs. Numerous diagnostic tests have been developed and are applied in two ways. The first is to confirm the diagnosis of JD in the presence of clinical signs. The second is for early detection of subclinically infected individuals in an attempt to remove infected (i.e. exposed but not shedding) and

infectious (i.e. exposed and shedding) animals from the herd to avoid further disease transmission and environmental contamination with MAP. In pathology, the gold standard test, the test that is considered the most accurate to diagnose the infection, is tissue culture of ileum tissue and adjacent lymph nodes with subsequent polymerase chain reaction (PCR) analysis to distinguish between MAP and other mycobacteria. The PCR uses the presence of the insertion segment

IS900 and potentially the DNA sequences F57, ISMav2, and ISMap02 to identify MAP (World

Organisation for Animal Health (OIE), 2012). For in vivo diagnostics, fecal culture with confirmative PCR analysis is often considered the gold standard test. However, culturing MAP is very difficult, requiring multiple decontamination steps and a long incubation period of 4 to 6 months (World Organisation for Animal Health (OIE), 2012). Direct fecal PCR, without prior culturing has been developed, but it is still not widely adopted (Clark et al., 2008). Diagnostic tests based on antibodies against the bacteria have been established, with the most common one being Enzyme Linked Immuno Sorbent Assay (ELISA) applied to blood serum or milk.

Estimates for sensitivity and specificity of common JD tests are presented in Table 1.1. The sensitivity of milk and serum ELISA, and direct fecal PCR, increases as cows increase the number of colony forming units (CFU) of MAP shed in the feces (Hendrick et al., 2005b; Clark et al., 2008). Furthermore, a high S:P ratio or high OD for serum and milk ELISA, respectively, is indicative of fecal shedding (Hendrick et al., 2005b). Due to the slow disease progression, the highest probability of testing positive is found in cows aging 2.5 to 4.5 and 2.5 to 5.5 years for

ELISA and fecal culture, respectively (Nielsen and Ersbøll, 2006). However, the proportion of heifers <2 years old that started fecal shedding increases as the within-herd JD prevalence increases (Weber et al., 2010).

Nonetheless, there are still major deficits in diagnosing infected animals in vivo and many infectious animals may remain undetected. Therefore, most JD control programs focus on

prevention of disease transmission rather than on test and cull strategies. A simulation study by

Groenendaal et al. (2003) showed that a JD control strategy involving management improvement in the calving and calf rearing area (<6 months) will be more successful in reducing the JD prevalence than testing for JD and culling test positive animals only.

Table 1.1 Sensitivity and specificity of diagnostic tests for JD Test Comparison (Gold Standard) Sensitivity [%] Specificity [%] Reference Milk ELISA Fecal culture 61.1 (48.9-72.4) 94.7a Hendrick et al., 2005b 28.9 99.7 Collins et al., 2005 Serum ELISA Fecal culture 73.6 (61.9-83.3) 87.5 a Hendrick et al., 2005b 16.7 (4.5–28.8) 97.1 (96.0–98.1) McKenna et al., 2005c 13.9 (2.6–25.2) 95.9 (94.6–97.2) McKenna et al., 2005 c 27.8 (13.1–42.4) 90.1 (88.2–92.0) McKenna et al., 2005 c 31.3 (20.6-43.8) 97.8 (94.5-99.4) Clark et al., 2008 28.9 95.3 Collins et al., 2005 28.4 99.7 Collins et al., 2005 28.0 100.0 Collins et al., 2005 44.5 84.9 Collins et al., 2005 57 98.9 Milner et al., 1990 Serum ELISA Tissue culture 8.8 (4.4–13.1) 97.6 (96.6–98.6) McKenna et al., 2005 6.9 (3.0–10.8) 96.0 (94.7–97.4) McKenna et al., 2005 16.9 (11.0–22.7) 90.8 (88.8–92.7) McKenna et al., 2005 Fecal culture Repeated fecal culture 38 100 Whitlock et al., 2000 Direct fecal PCR Fecal culture 70.2 (57.7-80.7) 85.3 (79.3-90.1) Clark et al., 2008 Environmental Individual fecal culture 71.4 (49.2-86.5) 98.6 (94.8-99.6) Lavers et al., 2012 cultured 76.0 80.0a Lombard et al., 2006 Individual serum ELISA 76.3 58.8a Lombard et al., 2006 Individual milk ELISA 71.4 66.7a Lombard et al., 2006 Bulk tank milk Sensitivity: Individual fecal 97.1 (83-100)b 83.3 (74-90)b Nielsen et al., 2000 ELISAd culture from Danish herds; Specificity: bulk tank samples from Norwegian herds considered to be negative a Calculated based on data presented b Based on a cut-off (OD) of 0.02 c Different estimates of the same type of test by the same author indicate the use of different tests (e.g. different companies) or different test characteristics (e.g. different cut points) d Herd-level test

1.4.3 Johne’s disease prevalence and control in Canada To decrease the prevalence and economic consequences of JD in Canadian dairy herds, the dairy industry (Dairy Farmers of Canada, Canadian Cattlemen’s Association, and the Canadian Animal

Health Coalition), governments, and veterinary schools collaboratively created the Canadian

Johne’s Disease Initiative (CJDI). The focus of this initiative is on education and awareness about

JD, coordination of provincial working groups, and coordination and facilitation of JD research activities. The JD programs are administered provincially and most provinces have established a

JD control program (Barker et al. 2012). The duration of the programs, the budgets, as well as participation differ greatly by province and region (Table 1.2). It is noteworthy that the majority

(88%) of the Ontario program was funded by the dairy industry rather than the public.

Table 1.2 Description of Johne's disease control programs in Canadian provinces and regions (Ron Barker, unpublished) Region/Province Duration Budget Main Number of Number of sponsoring farms trained dairy body participating veterinarians (%) (%) Atlantic Canadaa 2011-2014 1,000,000 Public 459 (69) 49 (60) Quebec 2007-2014 1,600,000 Public 1362 (22) 161 (47) Ontario 2010-2013 2,440,000 Industry 2339 (58) 246 (>95) Manitoba 2010-2011 100,000 Public 200 (57)b 20b Saskatchewan 2012-2013 125,000 Public 20 (12) 10b Alberta 2010-2013 1,040,000 Public 350 (61) 78 (95) British Columbia 2009-2012 250,000 Public 30 (6) 11 (50) Canada 2007-2014 6,600,000 Public 4759 (>35) 575 (>60) a New Brunswick, Nova Scotia, Prince Edward Island b approximate

When delivering the program at the farm level, trained veterinarians gather the herd history and evaluate the animal management practices using a standardized risk assessment instrument, and then make recommendations regarding herd testing. The herd owner and the veterinarian decide on a plan to implement best management practices for JD control. The tool used for this process is a Risk Assessment and Management Plan (RAMP). While the RAMP was adopted more uniformly across the provincial programs, the testing methods differ substantially. Some

programs utilize environmental testing alone (Alberta Johne’s Disease Initiative) or in combination with individual animal testing (Atlantic Johne’s Disease Initiative). Other programs use only individual testing (Ontario Johne’s Disease Education and Management Assistance

Program, Quebec Voluntary Paratuberculosis Prevention and Control Program). The tests include fecal culture and fecal PCR, as well as milk and serum ELISA (Barker et al. 2012). Prevalence estimates for the provinces are presented in Table 1.3.

Table 1.3 Johne's disease dairy herd-level prevalence in Canadian provinces Province Test Herd-level Reference prevalence [%] New Brunswick Serum ELISA 43.3a (24.5-62.2) VanLeeuwen et al., 2001 Nova Scotia Serum ELISA 53.3a (34.4-72.2) VanLeeuwen et al., 2001 Prince Edward Island Serum ELISA 33.3a (15.4-51.2) VanLeeuwen et al., 2001 Atlantic provinces d Environmental 20 Atlantic Johne’s Disease culture Initiative, 2012c Quebec Serum ELISA 42.5 Ministère de l’Agriculture, des Pêcheries et de l’Alimentation (MAPAQ), 2002c Ontario Serum ELISA 58a (44-72) Hendrick et al., 2005a Milk ELISA 34a (21-47) Hendrick et al., 2005a Milk or Serum 26 Ontario Johne’s ELISA Education and Management Assistance Program, 2013c Manitoba Serum ELISA 68.4a (52.5-84.2) VanLeeuwen et al., 2006 Saskatchewan Serum ELISA 43.3a (27.4-59.3) VanLeeuwen et al., 2005 Alberta Serum ELISA 40.0 ± 13.6b Sorensen et al., 2003 Serum ELISA 70.2a (53.7-86.6) Scott et al., 2006 Environmental 70 Alberta Johne’s Disease culture Initiative, 2014c a Herd considered positive if 1 or more cows tested positive b Herd considered positive if 2 or more cows tested positive c Provincial or regional Johne’s disease program website d Atlantic provinces include Nova Scotia, New Brunswick, Prince Edward Island, and Newfoundland and Labrador

1.5 JOHNE’S DISEASE PREVALENCE ON ORGANIC DAIRY FARMS

Very few researchers have investigated Johne’s disease prevalence on organic dairy farms. Three of them (Kijlstra, 2005; Ramanantoanina et al., 2012; Sorge, 2014) are published as conference proceedings that are usually limited by space provided and are therefore lacking in detail.

One report from The Netherlands states that the Johne’s disease prevalence on organic farms was similar to the prevalence on conventional farms (Kijlstra, 2005). In 2003, the authors collected blood from almost 3,700 animals that were older than 36 months from 76 organic farms and tested these samples with a serum ELISA. Using a concurrent risk assessment, they reported higher risk scores (higher risk for disease transmission) for the biosecurity, calving, and pre- and post-weaned calf management areas on organic farms compared to measures from previous studies. The authors reported an individual serum prevalence of 1.2% and 1.7%, and a herd level prevalence of 36% and 39% on organic and conventional farms, respectively. While they found more high risk management practices on organic farms, they concluded that there was no difference in serum ELISA prevalence (Kijlstra, 2005). The study does not comment on herd size in their report and it is therefore difficult to put the findings into perspective.

A survey from Quebec, Canada, studied a regionally stratified random sample of 60 organic dairy herds (Ramanantoanina et al., 2012). They conducted a serum ELISA on 30 cows in each of those herds. The authors found an individual serum prevalence of 0.8% (CI: 0.0-1.3%) and a herd level prevalence of 20.3% (CI: 10.0- 32.8%). It was concluded that MAP was probably less of an important disease on organic than on conventional farms based on a comparison with historical estimates of JD prevalence in Quebec (Ramanantoanina et al., 2012). In this study, it is unclear how the herd size and other characteristics of the sample herds compare to those of conventional

Quebec farms. It is also unclear how the historical estimates were derived (e. g. fecal culture, milk ELISA, or environmental sample).

A concurrent study on organic and conventional dairy farms in Minnesota, USA, used environmental samples to determine MAP infection status (Sorge, 2014). The author reported a lower MAP herd-level prevalence on organic and small (<200 cows) conventional dairy farms compared to large (≥ 200 cows) conventional dairy farms (43%, 47%, and 92%, respectively).

Similarly, Zwald et al. (2004) stated that Johne’s disease was more commonly reported by producers on conventional than organic dairies based on a management questionnaire of a sample of farms from Michigan, Minnesota, New York, and Wisconsin, USA. However, the conventional herds were larger than the organic herds in this study. Commonly, larger herds have a higher JD prevalence than smaller herds (Muskens et al., 2003; Wells and Wagner, 2000).

Therefore, comparisons cannot easily be made without adjustments for herd size.

Despite their limitations, these studies give an indication of how Johne’s disease prevalence compares between organic and conventional farms. However, they are of limited value in understanding the relationship between organic farming and Johne’s disease. In the following sections, a comparison between organic practices and Johne’s prevention practices is presented.

Where appropriate, links to additional infectious diseases are given to further support critical evaluation of organic management practices.

1.6 ASSOCIATION BETWEEN JOHNE’S DISEASE AND ORGANIC DAIRY

FARMING

1.6.1 Farm structure During the ongoing process of intensification of dairy farming and agriculture as a whole, dairy farms are required to grow or, if they are unable to do so, increase farm income by other means to avoid exiting the industry. While there are many other reasons why producers convert to organic farming, some smaller farms might see an opportunity to avoid the consolidation process and increase farm income by converting to organic milk production. More recent reports demonstrate 17

that organic farms tend to be smaller than conventional farms (Rozzi et al., 2007, Roberts et al.,

2008, Sato et al., 2005, Zwald et al., 2004; Figure 1.3). Small farms have been shown to have decreased odds of being JD positive (Muskens et al., 2003; Wells and Wagner, 2000) and, thus, smaller organic farms might have a reduced JD herd level prevalence compared to larger conventional farms.

Figure 1.3 Average herd size in organic and conventional Canadian dairy herds (data from Stonehouse et al., 2001, Rozzi et al., 2007, Roberts et al., 2008, Pieper et al., 2013)

On the other hand, compared to larger specialized farms (e.g. only milk production), smaller organic farms might still practice mixed farming (e.g. presence of other livestock species) for commercial purposes or own use. This might increase the risk of transmission of diseases that are shared among other species and dairy cattle. A clear separation between animals of different species is not always possible or desirable. For example, some farmers might let chickens follow cows on pasture and pick through the dung pats to search for food (e. g. fly maggots and larvae).

For parasite control, it is a common recommendation to rotate pasture use among different

species (e. g. horses and cattle). Based on superstition, some farmers might house a goat in the dairy cattle barn in the belief that it will prevent abortions (Foster, 1917). One might speculate that species that could harbor MAP but might not get tested for JD, could routinely come in direct or indirect contact with the cattle and could, therefore, transmit MAP to them.

A common claim in the organic industry is that because of a reduced milk yield, regular pasture access and higher forage content in the overall ration, as well as breeding for longevity and fitness, organically raised animals have a lower disease incidence and greater longevity than conventionally raised dairy animals. This is the foundation for the belief that organic farms have older cows than conventional farms. Stiglbauer et al. (2013) reported an average lactation number of 2.6 and 2.3 lactations, and an average proportion of 1st lactation animals of 31.6% and 37.3% on organic and conventional farms, respectively. Similarly, Hardeng and Edge (2001) reported a significant difference between the average lactation number of 2.97 on organic compared to 2.35 on conventional Norwegian dairy farms. Older cows are thought to be more profitable than younger cows because of increasing milk yield in higher lactations. However, older cows are also associated with an increase in health problems such as clinical mastitis, elevated SCC, milk fever, ketosis or JD (Hardeng and Edge, 2001, Sorge et al., 2011). JD is also more likely to be detected in older animals (Jubb et al., 2004). According to Tiwari et al. (2009), the odds of a farm being

JD negative decreases with increasing mean lactation number (OR=0.2, 95% CI: 0.1-0.8).

Therefore, a potentially increased JD prevalence on organic farms might be caused by an increased transmission of MAP by older cows and the greater likelihood of detection of their infection in these older animals. On the other hand it is unclear whether those biologically small differences will really lead to differences in disease prevalence or an increased risk of disease transmission.

1.6.2 Crop management Organic farmers try to maintain or increase soil health and fertility through the use of long crop rotations, cover crops, and utilization of manure and compost (Sholubi et al., 1997). Organic farms aim to preserve or increase organic matter in the soil which helps to retain water and nutrients in the soil and harbors microorganisms beneficial for plant health and performance

(Lynch, 2009). However, increasing soil organic matter has also been associated with increased survival of ovine MAP on pastures in Australia (Dhand et al., 2009). This could indicate an indirectly increased risk for MAP survival and transmission on organic dairy farms, but it remains unclear how bovine MAP behaves in soils with different organic content. Besides soil organic matter, other nutrients such as K, N and P might change under organic management

(Roberts et al., 2008). The effect of those changes on the survival of MAP on pastures or fields is also unclear. Furthermore, while the detection of viable MAP is greatly reduced in thermophilic compost compared to liquid manure, MAP DNA can be detected in compost up to 56 days and liquid manure up to 175 days (Grewal et al., 2006). Composting has also been shown to be ineffective for decontamination of carcasses infected with MAP, even after 250 days (Tkachuk et al., 2013). This indicates that proper thermophilic composting could reduce the danger of recycling MAP on the farm, but only if contaminated carcasses are disposed otherwise.

1.6.3 Nutrition Organic farmers feed less grain than conventional farmers (Richert et al., 2013) and consequently milk production is lower than on conventional farms (Richert et al., 2013; Zwald et al., 2004). In an attempt to decrease off-farm inputs and because of high purchase prices for certified organic feed, organic farmers tend to use fewer varieties of purchased feeds and fewer farms use any purchased feed (Zwald et al., 2004). This could make it more difficult to formulate nutrient dense, balanced rations with homegrown feed.

Grazing is a requirement of organic dairy production (Government of Canada, 2006). Sun exposure during grazing could, through increased vitamin D production in the skin and therefore calcium mobilization, help to reduce milk fever at calving. However, besides its importance in the calcium homeostasis, vitamin D plays a crucial role in immune functions and vitamin D deficiency has been associated with disease susceptibility (Di Rosa et al., 2011). It has recently been shown that the vitamin D concentration in cows with positive JD serum ELISA status is lower than in cows with a negative status (Sorge et al., 2013). Although the aforementioned authors do not indicate a causal relationship, research needs to be done to determine if pasturing cattle might be a strategy to suppress spread of MAP infection and clinical signs of JD through the mechanism of increased vitamin D availability.

Appropriate pasture management is complex and the farmer needs to assure that the animals have access to feed and water, that dry cows have ideal body condition at calving, that milking cows and young stock have access to enough high-quality feed to sustain milk production or growth, respectively, that the parasite burden is kept at a minimum, and that the animals have shelter.

Farmers also need to assure that the pastures are properly fenced so that the livestock are kept in the pastures, and predators and other wildlife that might share pathogens are kept out of the pastures. For example, wild ruminants and other wild non-ruminants (e.g. rabbits) can harbor and shed MAP (Machackova et al., 2004; Nugent et al., 2011; Judge et al, 2005). The presence of deer on the farm has been associated with increased odds of being JD positive (Cetinkaya et al,

1997). However, according to a study by Zwald et al. (2004), there is no difference in farmer reported contact rates with other farmed or wild animals between conventional and organic dairy farms. Nevertheless, it remains unclear how well rabbits or deer can be kept off the pastures at any time during the year. It is also unknown how long MAP shed by those wild animals can

survive on the pasture and if this poses a significant source of new infection for organic or pastured dairy cattle.

To sustain low parasite infestation in the herd and achieve sufficient, high-quality feed intake, farmers rotate their pastures regularly. Only a few researchers have investigated the survival of

MAP on pasture; it was found to survive up to 4 months (Machackova et al., 2004) in the open and up to 55 weeks in the fully shaded areas (Whittington et al., 2004). However, one can assume that a significant decontamination of the pastures is achieved within a few weeks. It was suggested that grazing with unsusceptible species or cutting could hasten decontamination due to decrease of shaded soil areas (Whittington et al., 2004). As common pasture rotations return cattle to potentially contaminated fields within 4 to 6 weeks, it is unlikely that pasture rotation alone will provide an effective measure against JD transmission. However, it is also doubtful that this management practice will increase JD transmission except if calves or heifers are rotated after adult cows as is commonly recommended for parasite control (Thamsborg et al., 1999).

Zwald et al. (2004) and Kijlstra (2005) reported that organic farms use surface water more frequently than conventional farms as drinking water for livestock. This might pose the risk of spread of diseases that are transmitted by organisms that survive in the water or in the moisture around the water accesses, such as internal parasites (e.g. liver fluke). Furthermore, as previously mentioned, MAP can survive up to 48 and 36 weeks in dam water or shaded and semi-exposed areas, respectively. The survival in the sediment was 12 to 24 weeks longer than in the water column. In this study, survival of MAP in soil and fecal material in the shade was only 12 weeks

(Whittington et al., 2005). However, it is unclear, how those experimental results translate to

MAP exposure from ponds and streams where cows have access. It is also unknown if Ontario organic dairy farmers use surface water as drinking water for their livestock.

In the wintertime, cows on organic farms are required to get regular outdoor access (Government of Canada, 2006) often provided through a winter paddock or exercise lot. Research from the

USA showed that the utilization of an exercise lot was associated with a three-fold increase in the odds of a herd being JD ELISA positive (Johnson-Ifearulundu and Kaneene, 1998). Although the definition of an exercise lot might vary from farm to farm, this indicates that the management

(e.g. cow density and cleanliness) of the exercise lot is insufficient on the majority of farms to prevent further spread of the disease.

1.6.4 Calving and dairy heifer management For JD control it is recommended to let cows calve in a clean calving area, separate from other cows rather than in group calving areas (Wraight et al., 2000). Organic farms have been reported to less often have a separate maternity area than conventional farms (Zwald et al., 2004, Kijlstra,

2005) and might therefore be at an increased risk of transmitting JD. Additionally, many organic farms let cows calve on pasture, where the animal density may be lower, but where access to other cows is possible. Furthermore, calves born on pasture would be more likely to suckle the first colostrum rather than be bottle or tube fed. To date, no research has been done to investigate the risk of JD transmission when cows calve on pasture.

According to the COS, calves have to be fed with natural milk within the first day of life

(Government of Canada, 2006). Nursing, in terms of using foster cows or just the own dam, is not a requirement in organic farming in Canada; however, it may be practiced among organic producers. In a study from The Netherlands, organic dairy producers were reported to leave the calves with the dam more often and feed artificial milk replacer less often (Kijlstra, 2005). In conventional dairy herds, it is generally recommended that calves are removed from the dam quickly after calving and that cleanly milked colostrum from the calf’s own dam, a JD test negative cow, or artificial colostrum replacer is given to the calf in sufficient amount soon after

birth (Wraight et al., 2000). Therefore, farms that let the calves suckle the dam could have an increased risk of JD transmission. On the other hand, calves fed colostrum from multiple cows and not just their own dam, are reported to have an odds ratio of 1.2 of testing JD positive compared to calves that received colostrum from their own dam only (Nielsen et al., 2008). It remains unclear if the risk might be lower when the calf is suckling a single test negative cow only rather than receiving a mixture of colostrum from recently calved cows. In the latter situation, an infected but still undiagnosed cow might be shedding MAP and could potentially infect multiple calves in a birth cohort receiving her colostrum or milk, whereas in the former situation only one calf or few calves get infected. In addition, if a nurse cow gets diagnosed with

JD later, her offspring or the calve(s) that received her milk and was exposed to her manure can easily be identified and removed from the herd if appropriate records exist on the farm.

Nevertheless, in the study by Nielsen et al. (2008), calves that suckled a foster cow had 2.0 times the odds of testing positive compared to calves fed milk replacer. This indicates that the management of foster cows might be still insufficient to prevent JD transmission.

Feeding high SCC or non-saleable milk is more common on organic than conventional farms in

The Netherlands (Kijlstra, 2005). This might be due to the imposition of longer withdrawal times for therapeutic treatments on organic compared to conventional farms. Although this practice is prohibited by the COS (Government of Canada, 2006), after the infection has cleared, farmers might use milk from treated cows to feed the calves and save saleable milk to increase their income. This practice could increase JD transmission to the calves as there seems to be a strong positive association between JD milk ELISA antibodies and high SCC (Baptista et al., 2008) and farmers would be more likely to feed high risk milk to susceptible calves. Tavornpanich et al.

(2008) reported that feeding non-saleable milk was a risk factor for having a high JD ELISA prevalence in the herd.

According to a study by Zwald et al. (2004), organic farms group feed pre-weaned calves more often than conventional farms. In Canada, organic farms that do not use nurse cows can house their calves either together or individually in hutches with access to a little yard until three months of age, provided they are not tethered, and the calves can see, hear and smell other calves.

After weaning, calves have to be housed together (Government of Canada, 2006). This increased contact between calves is likely positive for the development of the calves’ social skills (Duve et al., 2012) but might pose a risk for infectious disease transmission among calves. For JD control, it is recommended to isolate calves from each other until 30 days of age (Wraight et al., 2000).

Mitchell et al. (2011) showed that most MAP infected calves enter an early shedding period in which they can transmit the bacteria to other calves (van Roermund et al., 2007). Group housing replacement heifers might therefore increase the risk for JD transmission on organic and conventional dairy farms. In contrast, Marcé et al. (2011) demonstrated in their modelling study that group housing of calves did not influence MAP transmission in a herd.

Depending on the season, replacement calves that are greater than nine months old must have pasture access (Government of Canada, 2006). This will again pose an increased risk for parasite infestation and suboptimal nutrition during that time. Heifers that are co-housed or rotated with adult cows might also have an increased risk of getting infected with MAP. In a meta-analysis, it was specified that about 50% of heifers inoculated at 6 to 12 months of age develop lesions indicative of bovine JD (Windsor and Whittington, 2010), showing that infection at those ages still occurs. According to Kijlstra (2005), Dutch organic farmers keep their calves more often than conventional farms on pastures that have been used by adult cattle or goats, or that have been fertilized with cattle or goat manure earlier that season. On the other hand, Marcé et al.

(2011) showed that the longer contact between calves and cows could be delayed, the more the prevalence can be decreased over time in a JD control program. Therefore, co-housing or rotating

calves after cows might make JD control on organic farms more difficult than on conventional, non-grazing farms.

1.6.5 Veterinary treatments As preventive or metaphylactic treatment of animals is commonly banned in organic farming, it becomes challenging to manage parasite burden in pastured animals. Specific parasiticides are allowed in cases where an individual animal has a proven infestation with parasites and is unable to cope with this situation on its own (Government of Canada, 2006). Monensin is an example of a coccidiostatic and growth promoting medication and is commonly fed on conventional farms, but is not allowed in organic production. Feeding monensin has been shown to marginally reduce the amount of MAP shed in a clinical trial (Hendrick et al., 2006b) but was also associated with reduced odds of testing milk ELISA positive in an observational study (Hendrick et al., 2006a).

In a comparative study by Zwald et al. (2004), 22% of conventional herds used monensin as a feed additive for pre-weaned and weaned heifers whereas none of the organic herds did. As a result, organic farms might have an increased prevalence of cows testing positive for JD.

In the case where an animal gets sick, organic farming regulations prefer alternative treatment options (homeopathics, herbal medicine) over chemical allopathic veterinary drugs (Government of Canada, 2006). In western Ontario, Canada, homeopathic remedies are commonly used either in conjunction with chemical allopathic veterinary drugs or alone (Sholubi et al., 1997).

Moreover, the use of homeopathy is widely disseminated (10 - 42% of farms) among organic farmers for mastitis therapy as reported from three European countries (Wagenaar et al., 2011).

Multiple problems arise from the use of alternative treatments. In some cases, the use of alternative treatments is not in agreement with national drug and food safety regulations as most alternative treatment options are not licensed for veterinary usage for the desired indication.

Therefore, with the exception of homeopathic remedies, it would be required that veterinarians

prescribe them if they are to be used in dairy cattle. However, for the majority of alternative treatment options, clinical studies on efficacy, safety or withdrawal times are non-existent

(Ruegg, 2009) and consequently, practitioners might hesitate or are simply unable to legally prescribe them all together.

Veterinarians’ lack of support for those alternative treatments might force the farmer to seek advice from other health professionals, lay or sales people or other farmers. It has been reported that veterinarians are not utilized as much on organic as on conventional farms (Stonehouse et al.,

2001). Organic farms have been reported to have fewer veterinary visits per year, fewer routine veterinary visits, are less likely to call for emergencies, and fewer farms utilize preventive vaccination, consult with a nutritionist, or enroll in regular milk recording compared to conventional farms (Richert et al., 2013). However, it should be noted that veterinary usage is more associated to intensity of management rather than organic or conventional status (Richert et al., 2013). Organic farmers that do not regularly utilize those external inputs might also be less likely to participate in a disease prevention program (e.g. JD control) led by veterinarians. In the comparison of participants and non-participants in the voluntary Ontario, Canada, JD program, more progressive farmers with better management were more likely to participate in the program.

Among the reasons cited for non-participation was the unwillingness to pay the veterinarian to do the risk assessment and a farm policy to not participate in any formal programs (Kelton et al.,

2012). Organic farmers might, therefore, be less likely to participate in a JD control program or seek advice for JD control from their veterinarians.

Furthermore, an organic farmer might be tempted to cull a sick animal early to avoid further disease transmission, high veterinary costs, or long withdrawal times for veterinary treatments.

Thus, organic farmers tend to have fewer consultations with external dairy professionals which might pose the risk that diseases such as mastitis or JD arise but remain undetected for a longer

time. On the other hand, culling, rather than trying to treat diseased animals, potentially reduces environmental contamination with disease pathogens. In a modeling study, decreasing the mean time clinical JD cows spent on the farm, decreased the JD prevalence over time (Marcé et al.,

2011).

Another favorable aspect of early culling of diseased animals is the selection for disease resistance. If through early culling diseased animals are less likely to reproduce, healthy animals will have a reproductive advantage over sick/susceptible animals and their genes would subsequently have a higher abundance within the population. This would potentially lead to offspring with higher disease resistance and reduced disease incidence over time. Thompson-

Crispi et al. (2012) showed that decreased disease incidence was associated with high adaptive immune response measured by a standardized test in an US Holstein dairy herd. Additionally,

Pinedo et al. (2009) demonstrated that increasing cellular immune response was associated with reduced odds of serum ELISA positivity. However, a Dutch modelling study by van Hulzen et al.

(2014) showed that responses to JD ELISA-negative selection alone are relatively small, potentially requiring centuries for JD to be eliminated from the dairy herd using this technique.

To date, no research has investigated the distribution of high immune responders in organic versus conventional dairy farms and it is unknown if the culling strategy and disease selection on organic farms will lead to a higher prevalence of high immune responders and, therefore, reduced

JD prevalence.

1.6.6 Biosecurity As mentioned above, organic farmers tend to have fewer consultations with external dairy professionals. With this strategy, organic farmers might try to limit the number of professionals with regular animal contact entering the farm. This might be a measure of biosecurity to avoid transmission of highly infectious diseases such as Bovine Viral Diarrhea (BVD), Infectious

Bovine Rhinotracheitis (IBR), Bovine Respiratory Disease (BRD), or winter dysentery, again, because treatment options are very limited. Similarly, as a measure of biosecurity and due to problems with sourcing certified organic inputs, organic farmers tend to raise their own replacement animals (Stiglbauer et al., 2013; Stonehouse et al., 2001) and limit purchase of expensive external inputs such as feed (Zwald et al., 2004; Stonehouse et al., 2001). It has been shown that farms that regularly purchase animals have a higher risk of being Johne’s positive and of having higher JD seroprevalence than herds that stay closed (Wells and Wagner, 2000, Chi et al, 2002; Tiwari et al., 2009). Therefore, organic farms might be less likely to introduce the disease into the herd by external inputs. Contrarily, Kijlstra (2005) reported that a sample of

Dutch organic farms had less stringent biosecurity measures than conventional farms, such that they less often provide separate farm clothes for visitors, more often purchase animals from farms with unknown paratuberculosis status, and more often use manure from other farms compared to conventional farms. To date, it is unknown how those biosecurity measures on Canadian organic and conventional farms compare to each other.

1.6.7 Breeding strategy The limitation to purchasing external breeding stock does not always extend to the breeding bull.

Organic regulations in Canada state a preference for natural breeding but allow for artificial insemination (Government of Canada, 2006). As stated above, organic farms more often keep a breeding bull on the farm and this bull is sometimes purchased from other farmers to reduce the risk of inbreeding. If organically-raised bulls are unavailable, conventionally-raised bulls can be used instead (Government of Canada, 2006). Therefore, bulls could be purchased from larger conventional herds or the auction market. Larger farms are more likely to have a positive JD herd status than smaller herds (Muskens et al., 2003; Wells and Wagner, 2000) and herd additions from those sources might cause a higher risk of introducing the disease into the herd.

Besides the source of the addition, a breeding bull poses a different risk for disease transmission that is often ignored. After raising or purchasing a breeding bull, he is often co-housed (on pasture or in the barn) with dairy replacement heifers to avoid the extra work of watching for heat and artificial insemination. In a large US based study, 66% of organic farms compared to only

44% of non-grazing conventional farms were reported to use natural service for some or all replacement heifers (Stiglbauer et al., 2013). During that period, the heifers are usually 10 to 20 months old and still marginally susceptible to the oral infection with MAP (Mitchell et al., 2011,

Windsor and Whittington, 2010).

However, of particular concern are the previously mentioned reports about additional routes of disease transmission associated with the bull. Between 1970 and 2010 researchers have demonstrated that infected breeding bulls can shed MAP not only in the feces but also in the semen (Khol et al., 2010; Buergelt et al., 2004; Ayele et al. 2004; Larsen et al. 1981, Larsen and

Kopecky, 1970), while still maintaining acceptable semen quality. They shed the bacteria at low to moderate levels (102-105/ml) and shedding might be intermittent (Khol et al., 2010). MAP has also been found in male reproductive organs at pathological examination (Khol et al., 2010;

Glawischnig, et al., 2004).

On the other end of a potential venereal transmission route, Merkal et al. (1982) and Owen and

Thoen (1983) showed that infecting heifers and cows intrauterine with large doses of MAP is possible, leading to abortions, positive antibody reactions in inoculated animals, prolonged recovery of the organism from the uterus, and likely antibody or MAP positive calves. Although it has not been proven that breeding bulls infect cows or heifers through mating, it could be suspected that bulls could play a role in transmission of JD and that organic farms are therefore at an increased risk of introducing and transmitting JD in the herd.

Canadian organic farming regulations also recommend utilizing breeds or lines that are adapted to the specific environmental conditions (Government of Canada, 2006). The predominant breed on organic farms is Holstein; however, organic farms more often use cross breeds (Richert et al.,

2013) or other purebred breeds than conventional farms. Rozzi et al. (2007) reported that the use of minor or heritage breeds was associated with increased usage of breeding bulls. In Canada,

Channel Islands breeds (e. g. Jersey, Guernsey) have been shown to have increased odds of being

JD ELISA positive (Sorge et al., 2011). Lombard et al. (2006b) showed that non-Holstein breeds are more likely to test positive using serum or milk ELISA. Therefore, when trading with or breeding those breeds, organic farmers may be more likely to introduce and spread JD on their farm.

1.7 KNOWLEDGE GAPS

Most of the aforementioned potential associations between JD and organic farming are based on the review of the broader literature and not on empirical studies on organic dairy farms.

Therefore, there are many knowledge gaps and research needs. Addressing those needs will not only benefit the organic dairy farming community, but will also help smaller conventional dairy farms that employ similar management practices, to reduce the risk for JD transmission.

To more clearly understand the relationship between organic farming and MAP transmission, a concurrent comparison between the organic and conventional farming systems in terms of JD prevalence and risk factors is necessary, while at the same time accounting for herd size and farming intensity as major confounders. This will rule out biases introduced by using historical data for comparison or differences in herd size and intensity that might impose differences in management practices.

The relationship between organic dairy farmers and veterinarians seems to be challenged by differing views and requirements of each profession imposed by organic requirements.

Researchers and educators need to be able to understand and make targeted recommendations and adjustments to improve this relationship. Only a strong and trustful relationship will facilitate the delivery of animal health and welfare programs among the organic livestock industry and results in improvement of animal wellbeing and food safety for the consumer.

As shown above, a number of researchers have investigated the prevalence of MAP in reproductive organs of infected bulls. It is unclear whether a breeding bull on dairy farms poses a major risk for JD transmission. A survey of breeding bulls on commercial organic and conventional dairy farms would help to determine the prevalence of MAP infected individuals.

Strain typing infected offspring of JD-positive bulls might further help to demonstrate an association. Challenge trials utilizing similar MAP concentrations found in semen would be needed to show if intrauterine infection through mating by infected bulls is possible.

As pasturing is a requirement under organic regulations, farmers are unlikely to change this management practice. It has been demonstrated that MAP can survive on pasture for very long time and that MAP can be shed by wild animals that also may reside on pastures. Research needs to be done to determine the extent of JD transmission on pasture and how to reduce it. Further, it needs to be investigated whether calving on pasture poses an increased risk compared to calving in the barn.

Another practice that farmers are unlikely to change is letting calves nurse the cows. It is therefore first necessary to determine the risk of JD transmission by suckling versus other feeding practices (e. g. group feeding milk to calves). Secondly, it needs to be investigated how the disease transmission risk could be decreased (e. g. only use test negative cows) to benefit from the positive aspects of nursing.

1.8 CONCLUSIONS

Farming practices and regulations might directly or indirectly alter the risk for JD transmission on Canadian organic dairy farms. Ultimately, estimates of how much each of those factors will alter JD transmission risk on organic farms are unavailable and predictions about overall change in herd-level and within-herd prevalence are impossible. Farmers engaging in organic practices might need to critically review their management regarding possibilities of infectious disease transmission. They may also need to test the herd for JD more frequently and pay special attention to hygiene and biosecurity measures. More research is needed to address major knowledge gaps. Closing those gaps will not only benefit organic farms, but also smaller conventional farms that employ management practices similar to organic farms.

1.9 ACKNOWLEDGMENTS

This research was kindly funded by the Ontario Ministry of Agriculture and Food (OMAF)-

University of Guelph research partnership.

1.10 OBJECTIVES

The objectives of this study were to:

1) Describe the organic dairy farming community in Ontario

2) Evaluate the Risk Assessment and Management Plan (RAMP) used for JD prevention

and control in Ontario

3) Compare prevalence, risk factors, and recommendations for JD prevention for organic

and conventional farms in Ontario

4) Determine barriers and opportunities for organic dairy farms to JD prevention and

control

5) Revise the RAMP to better suit the breadth of farms in Ontario

1.11 REFERENCES

Agriculture and Agri-Food Canada (AAFC). 2012. Organic Dairy Industry in Canada. 2012

Edition. Accessed January 5, 2014. http://www.dairyinfo.gc.ca/pdf/organic_profile_eng.pdf

Agriculture and Agri-Food Canada (AAFC). 2009. Certified Organic Production Statistics for

Canada. Accessed January 15, 2014. http://www.agr.gc.ca/eng/industry-markets-and- trade/statistics-and-market-information/by-product-sector/organic-products/organic-production- canadian-industry/certified-organic-production-statistics-for-canada-

2009/?id=1312385802597#a3

Alberta Johne’s Disease Initiative. 2014. Johne’s Disease-What’s the Scoop?. Accessed April 24,

2014. http://www.albertajohnes.ca/JDabout/MT_1401a.pdf

Atlantic Johne’s Disease Initiative. 2012. Johne’s Control Program Reaches Milestone. Accessed

April 24, 2014. http://www.atlanticjohnes.ca/files/documents/InsertSpring2012V1.pdf

Ayele, W. Y., M. Bartos, P. Svastova and I. Pavlik. 2004. Distribution of Mycobacterium avium subsp. paratuberculosis in Organs of Naturally Infected Bull-Calves and Breeding Bulls.

Veterinary Microbiology 103: 209-217.

Baptista, F. M., S. S. Nielsen, and N. Toft. 2008. Association Between the Presence of

Antibodies to Mycobacterium avium subsp. paratuberculosis and Somatic Cell Count. Journal of

Dairy Science 91(1):109-118.

Barker, R. A., H. W. Barkema, G. Fecteau, G. P. Keefe, D. F. Kelton. 2012. Johne’s Disease

Control in Canada – Coordinated Nationally – Delivered Provincially. Proceedings of the 3rd

ParaTB Forum. Sydney, Australia, February 4th, 2012.

Buergelt, C. D., G. A. Donovan and J. E. Williams. 2004. Identification of Mycobacterium avium subspecies paratuberculosis by Polymerase Chain Reaction in Blood and Semen of a Bull with

Clinical Paratuberculosis. International Journal of Applied Research in Veterinary Medicine 2:

130-134.

Cetinkaya, B., H. M. Erdogan, and K. L. Morgan. 1997. Relationships between the Presence of

Johne's Disease and Farm and Management Factors in Dairy Cattle in England. Prev Vet Med

32(3-4):253-266

Chi, J., J. A. VanLeeuwen, A. Weersink, and G. P. Keefe. 2002. Management Factors Related to

Seroprevalences to Bovine Viral-Diarrhoea Virus, Bovine-Leukosis Virus, Mycobacterium avium subspecies paratuberculosis, and Neospora caninum in Dairy Herds in the Canadian Maritimes.

Preventive Veterinary Medicine 55(1):57-68.

Clark, D. L., Jr, J. J. Koziczkowski, R. P. Radcliff, R. A. Carlson, and J. L. E. Ellingson. 2008.

Detection of Mycobacterium avium subspecies paratuberculosis: Comparing Fecal Culture

Versus Serum Enzyme-Linked Immunosorbent Assay and Direct Fecal Polymerase Chain

Reaction. Journal of Dairy Science 91(7):2620-2620-2627.

Collins, M. T., S. J. Wells, K. R. Petrini, J. E. Collins, R. D. Schultz, and R. H. Whitlock. 2005.

Evaluation of Five Antibody Detection Tests for Diagnosis of Bovine Paratuberculosis. Clin

Diagn Lab Immunol 12(6):685-692.

Cranfield, J., S. Henson and J. Holliday. 2010. The Motives, Benefits, and Problems of

Conversion to Organic Production. Agriculture and Human Values 27: 291-306.

Dhand, N. K., J. Eppleston, R. J. Whittington, and J.-A. L. M. L. Toribio. 2009. Association of

Farm Soil Characteristics with Ovine Johne's Disease in Australia. Preventive Veterinary

Medicine 89(1-2):110-120.

Diéguez, F. J., I. Arnaiz, M. L. Sanjuán, M. J. Vilar, and E. Yus. 2008. Management Practices

Associated with Mycobacterium avium subspecies paratuberculosis Infection and the Effects of the Infection on Dairy Herds. Veterinary Record 162(19):614-617.

Di Rosa, M., M. Malaguarnera, F. Nicoletti, and L. Malaguarnera. 2011. Vitamin D3: a Helpful

Immuno-Modulator. Immunology 134(2):123-139.

Duve, L. R., D. M. Weary, U. Halekoh, and M. B. Jensen. 2012. The Effects of Social Contact and Milk Allowance on Responses to Handling, Play, and Social Behavior in Young Dairy

Calves. J Dairy Sci 95(11):6571-6581.

Foster, J. J. 1917. Goat and Cows. Folklore 28(4):451-451.

Glawischnig, W., M. Awad-Masalmeh, D. Khaschabi and M. Schoenbauer. 2004. [Detection of

Mycobacterium avium subsp. paratuberculosis from the Testicles of a Clinically Infected

Breeding Bull] article in German with English abstract. Berl. Muench. Tieraerztl. Wschr. 117:

136-139.

Grewal, S. K., S. Rajeev, S. Sreevatsan, and F. C. Michel, Jr. 2006. Persistence of

Mycobacterium avium subsp. paratuberculosis and Other Zoonotic Pathogens during Simulated

Composting, Manure Packing, and Liquid Storage of Dairy Manure. Appl Environ Microbiol

72(1):565-574.

Groenendaal, H., M. Nielen, and J. W. Hesselink. 2003. Development of the Dutch Johne's

Disease Control Program Supported by a Simulation Model. Preventive Veterinary Medicine

60(1):69-90.

Government of Canada. 2006. CAN/CGSB-32.310-2006 Organic Production Systems - General

Principles and Management Standards.

Government of Canada. 2006. CAN/CGSB-32.311-2006 Organic Production Systems - Permitted

Substances Lists.

Government of Canada. 2009. Organic Products Regulations, 2009 (SOR/2009-176).

Hardeng, F. and V. L. Edge. 2001. Mastitis, Ketosis, and Milk Fever in 31 Organic and 93

Conventional Norwegian Dairy Herds. Journal of Dairy Science 84(12):2673-2679.

Hendrick, S., T. Duffield, K. Leslie, K. Lissemore, M. Archambault, and D. Kelton. 2005a. The

Prevalence of Milk and Serum Antibodies to Mycobacterium avium subspecies paratuberculosis in Dairy Herds in Ontario. Can Vet J 46(12):1126-1129.

Hendrick, S. H., T. F. Duffield, D. F. Kelton, K. E. Leslie, K. D. Lissemore and M. Archambault.

2005b. Evaluation of Enzyme-Linked Immunosorbent Assays Performed on Milk and Serum

Samples for Detection of Paratuberculosis in Lactating Dairy Cows. Journal of the American

Veterinary Medical Association 226: 424-428.

Hendrick, S. H., T. F. Duffield, K. E. Leslie, K. D. Lissemore, M. Archambault, R. Bagg, P.

Dick, and D. F. Kelton. 2006a. Monensin Might Protect Ontario, Canada Dairy Cows from

Paratuberculosis Milk-ELISA Positivity. Preventive Veterinary Medicine 76(3-4):237-248.

Hendrick, S. H., D. F. Kelton, K. E. Leslie, K. D. Lissemore, M. Archambault, R. Bagg, P. Dick, and T. F. Duffield. 2006b. Efficacy of Monensin Sodium for the Reduction of Fecal Shedding of

Mycobacterium avium subsp. paratuberculosis in Infected Dairy Cattle. Preventive Veterinary

Medicine 75(3-4):206-220.

Hill, S. B. and R. J. MacRae. 1992. Organic Farming in Canada. Agriculture, Ecosystems &

Environment 39:71-84.

Johnson-Ifearulundu, Y. J. and J. B. Kaneene. 1998. Management-Related Risk Factors for M. paratuberculosis Infection in Michigan, USA, dairy herds. Prev Vet Med 37(1-4):41-54.

Jubb, T. F., E. S. G. Sergeant, A. P. L. Callinan, and J. W. Galvin. 2004. Estimate of the

Sensitivity of an ELISA Used to Detect Johne's Disease in Victorian Dairy Cattle Herds.

Australian Veterinary Journal 82(9):569-573.

Kelton, D., N. Perkins, A. Godkin, G. MacNaughton, R. Cantin, and K. Hand. 2012. Comparison of Participants and Non-Participants in a Voluntary Johne's Disease Control Program in Ontario,

Canada. in 11th International Colloquium on Paratuberculosis (11ICP), Sydney, Australia.

Khol, J. L. J., P. P. Kralik, I. I. Slana, V. V. Beran, C. C. Aurich, W. W. Baumgartner and I. I.

Pavlik. 2010. Consecutive Excretion of Mycobacterium avium subspecies paratuberculosis in

Semen of a Breeding Bull Compared to the Distribution in Feces, Tissue and Blood by IS900 and

F57 Quantitative Real-Time PCR and Culture Examinations. The Journal of veterinary medical science / the Japanese Society of Veterinary Science 72: 1283-1283-1288.

Kijlstra, A. 2005. No Difference in Paratuberculosis Seroprevalence Between Organic and

Conventional Dairy Herds in The Netherlands. In: Hovi, M.; Zastawny, J. und Padel, S. (Hrsg.)

Proceedings of the 3rd SAFO Workshop; Enhancing Animal Health Security and Food Safety in

Organic Livestock Production., University of Reading, Reading, UK, S. 51-56.

Larsen, A. B. and K. E. Kopecky. 1970. Mycobacterium paratuberculosis in Reproductive

Organs and Semen of Bulls. American Journal of Veterinary Research 31: 255-258.

Larsen, A. B., O. H. V. Stalheim, D. E. Hughes, L. H. Appell, W. D. Richards and E. M. Himes.

1981. Mycobacterium paratuberculosis in the Semen and Genital Organs of a Semen-Donor Bull.

Journal of the American Veterinary Medical Association 179: 169-171.

Lavers, C., G. P. Keefe, S. Mckenna, and M. Chaffer. 2012. Evaluation of Environmental Culture to Determine Herd Status for Johne's Disease on Atlantic Canadian Dairy Farms. Pages 52-53 in

11th International Colloquium on Paratuberculosis (11ICP), Sydney, Australia.

Lombard, J. E., B. A. Wagner, R. L. Smith, and B. J. McCluskey. 2006. Evaluation of

Environmental Sampling and Culture to Determine Mycobacterium avium subspecies paratuberculosis Distribution and Herd Infection Status on US Dairy Operations. Journal of

Dairy Science 89(11):4163-4171,4165.

Lombard, J. E., T. M. Byrem, B. A. Wagner, and B. J. McCluskey. 2006b. Comparison of Milk and Serum Enzyme-Linked Immunosorbent Assays for Diagnosis of Mycobacterium avium subspecies paratuberculosis Infection in Dairy Cattle. J Vet Diagn Invest 18(5):448-458.

Lynch, D. 2009. Environmental Impacts of Organic Agriculture: A Canadian Perspective.

Canadian Journal of Plant Science 89(4):621-628.

Machackova, M., P. Svastova, J. Lamka, I. Parmova, V. Liska, J. Smolik, O. A. Fischer, and I.

Pavlik. 2004. Paratuberculosis in Farmed and Free-Living Wild Ruminants in the Czech Republic

(1999-2001). Veterinary Microbiology 101(4):225-234.

Marcé, C., P. Ezanno, H. Seegers, D. U. Pfeiffer, and C. Fourichon. 2011. Within-Herd Contact

Structure and Transmission of Mycobacterium avium subspecies paratuberculosis in a

Persistently Infected Dairy Cattle Herd. Preventive Veterinary Medicine, SVEPM 2010, 2010

Society of Veterinary Epidemiology and Preventive Medicine conference 100(2):116-125.

McKenna, S. L. B., G. P. Keefe, H. W. Barkema, and D. C. Sockett. 2005. Evaluation of Three

ELISAs for Mycobacterium avium subsp. paratuberculosis Using Tissue and Fecal Culture as

Comparison Standards. Veterinary Microbiology 110:105-111.

Merkal, R. S., J. M. Miller, A. M. Hintz and J. H. Bryner. 1982. Intrauterine Inoculation of

Mycobacterium paratuberculosis into Guinea Pigs and Cattle. American Journal of Veterinary

Research 43: 676-678.

Milner, A. R., W. N. Mack, K. J. Coates, J. Hill, I. Gill, and P. Sheldrick. 1990. The Sensitivity and Specificity of a Modified ELISA for the Diagnosis of Johne's Disease from a Field Trial in

Cattle. Vet Microbiol 25(2-3):193-198.

Ministère de l’Agriculture, des Pêcheries et de l’Alimentation (MAPAQ). 2002. Paratuberculose.

Accessed April 24. 2014. http://www.mapaq.gouv.qc.ca/fr/Productions/santeanimale/maladiesanimales/ paratuberculose/Pages/paratuberculose.aspx

Mitchell, R. M., G. F. Medley, M. T. Collins, and Y. H. Schukken. 2011. A Meta-Analysis of the

Effect of Dose and Age at Exposure on Shedding of Mycobacterium avium subspecies paratuberculosis (MAP) in Experimentally Infected Calves and Cows. Epidemiol Infect 28:1-16.

Muskens, J., A. R. W. Elbers, H. J. Van Weering, and J. P. T. M. Noordhuizen. 2003. Herd

Management Practices Associated with Paratuberculosis Seroprevalence in Dutch Dairy Herds.

Journal of Veterinary Medicine, Series B 50(8):372-377.

Nebbia, P., P. Robino, E. Ferroglio, L. Rossi, G. Meneguz, and S. Rosati. 2000. Paratuberculosis in Red Deer (Cervus elaphus hippelaphus) in the Western Alps. Veterinary Research

Communications 24(7):435-435-443.

Nielsen, S. S. and A. K. Ersbøll. 2006. Age at Occurrence of Mycobacterium avium subspecies paratuberculosis in Naturally Infected Dairy Cows. Journal of Dairy Science 89(12):4557-4557-

4566.

Nielsen, S. S., S. M. Thamsborg, H. Houe, and V. Bitsch. 2000. Bulk-tank Milk ELISA

Antibodies for Estimating the Prevalence of Paratuberculosis in Danish Dairy Herds. Preventive

Veterinary Medicine 44(1-2):1-7.

Nugent, G., E. J. Whitford, J. C. Hunnam, P. R. Wilson, M. L. Cross, and G. W. de Lisle. 2011.

Mycobacterium avium subsp. paratuberculosis Infection in Wildlife on Three Deer Farms with a

History of Johne's Disease. New Zealand Veterinary Journal 59(6):293-298.

Owen, W. J. and C. O. Thoen. 1983. Experimental Exposure of Cattle to Mycobacterium paratuberculosis Orally and Intrauterine with Attempted Culture of the Organs and Detection of

Humoral Antibodies. Proceedings of the United States Animal Health Association 87: 570-581.

Ogini, Y. O., A. Clark and D. P. Stonehouse. 1999. Comparison of Organic and Conventional

Dairy Farms in Ontario. American Journal of Alternative Agriculture 14: 122-128.

Ontario Johne’s Education and Management Assistance Program. 2013. Program Results.

Accessed April 24, 2014. http://www.johnes.ca/program%20results.htm

Pieper, L., A. Godkin, U. Sorge, K. Lissemore, T. J. DeVries, and D. Kelton. 2013. Prevalence of

Mycobacterium avium subsp paratuberculosis (MAP) ELISA-Positive Cows and Assessment of

MAP Transmission Risk on Organic Dairy Farms in Ontario, Canada. Pages 156-157 in

Proceedings of the 46th Annual Conference of the American Association of Bovine Practitioners

(AABP), Milwaukee, Wisconsin.

Pinedo, P. J., G. A. Donovan, O. Rae, and J. De la Paz. 2009. Association Between

Paratuberculosis Infection and General Immune Status in Dairy Cattle. Page 127 in 10th

International Colloquium for Paratuberculosis, Minneapolis, Minnesota, USA.

Pithua, P., S. M. Godden, S. J. Wells, and M. J. Oakes. 2009. Efficacy of Feeding Plasma-

Derived Commercial Colostrum Replacer for the Prevention of Transmission of Mycobacterium avium subsp paratuberculosis in Holstein Calves. Journal of the American Veterinary Medical

Association 234(9):1167-1176.

Ramanantoanina, F., D. Francoz, G. Côté, O. Labrecque, A. C. Gagnon, G. Beauchamp, and G.

Fecteau. 2012. Seroprevalence of Mycobacterium avium subsp. paratuberculosis, Neospora caninum, Bovine Viral Diarrhea and Infectious Bovine Rhinotracheitis Viruses in Québec

Organic Herds. in XXVII. World Buiatrics Congress, Lisbon, Portugal.

Richert, R. M., K. M. Cicconi, M. J. Gamroth, Y. H. Schukken, K. E. Stiglbauer and P. L. Ruegg.

2013. Management Factors Associated with Veterinary Usage by Organic and Conventional

Dairy Farms. Journal of the American Veterinary Medical Association 242: 1732-1743.

Roberts, C. J., R. C. Martin, S. D. Juurlink, D. H. Lynch and R. P. Voroney. 2008. Nutrient

Budgets of Ontario Organic Dairy Farms. Canadian journal of soil science. - Revue Canadienne de la science du sol 88: 107-114.

Rolle, M. and A. Mayr. 2007. Medical Microbiology and Epidemiology [Medizinische

Mikrobiologie, Infektions- und Seuchenlehre]. Vol. 8. Enke, Stuttgart, Germany.

Rozzi, P., F. Miglior and K. J. Hand. 2007. A Total Merit Selection Index for Ontario Organic

Dairy Farmers. Journal of Dairy Science 90: 1584-1593.

Ruegg, P. L. 2009. Management of Mastitis on Organic and Conventional Dairy Farms. Journal of Animal Science 87: 43-55.

Scott, H. M., O. Sorensen, J. T. Wu, E. Y. Chow, K. Manninen, and J. A. VanLeeuwen. 2006.

Seroprevalence of Mycobacterium avium subspecies paratuberculosis, Neospora caninum,

Bovine Leukemia Virus, and Bovine Viral Diarrhea Virus Infection among Dairy Cattle and

Herds in Alberta and Agroecological Risk Factors Associated with Seropositivity. Can Vet J

47(10):981-991.

Sholubi, Y. O., D. P. Stonehouse and E. A. Clark. 1997. Profile of Organic Dairy Farming in

Ontario. American Journal of Alternative Agriculture 12: 133.

Sorensen, O., S. Rawluk, J. Wu, K. Manninen, and G. Ollis. 2003. Mycobacterium paratuberculosis in Dairy Herds in Alberta. Canadian Veterinary Journal 44:221-226.

Sorge, U. S. 2014. An Overview Over the Organic Dairy Industry in Minnesota. in 117th Annual

Meeting of the Minnesota Veterinary Medical Association, Minneapolis, MN, USA. February 6-

8, 2014.

Sorge, U. S., K. Lissemore, A. Godkin, S. Hendrick, S. Wells and D. Kelton. 2011. Associations

Between Paratuberculosis Milk ELISA Result, Milk Production, and Breed in Canadian Dairy

Cows. Journal of Dairy Science 94: 754-761.

Sorge, U. S., S. Kurnick, and S. Sreevatsan. 2013. Detection of Mycobacterium avium subspecies paratuberculosis in the Saliva of Dairy Cows: A Pilot Study. Veterinary Microbiology 164(3-

4):383-386.

Sorge, U. S., T. Molitor, J. Linn, D. Gallaher, and S. W. Wells. 2013. Cow-Level Association

Between Serum 25-hydroxyvitamin D Concentration and Mycobacterium avium subspecies paratuberculosis Antibody Seropositivity: A Pilot Study. Journal of Dairy Science 96(2):1030-

1037.

Stiglbauer, K. E., K. M. Cicconi-Hogan, R. Richert, Y. H. Schukken, P. L. Ruegg and M.

Gamroth. 2013. Assessment of Herd Management on Organic and Conventional Dairy Farms in the United States. Journal of Dairy Science 96: 1290-1300.

Stonehouse, D. P., E. A. Clark and Y. A. Ogini. 2001. Organic and Conventional Dairy Farm

Comparisons in Ontario, Canada. Biological Agriculture & Horticulture 19: 115-125.

Streeter, R. N., G. F. Hoffsis, S. Bech-Nielsen, W. P. Shulaw, and D. M. Rings. 1995. Isolation of

Mycobacterium paratuberculosis from Colostrum and Milk of Subclinically Infected Cows.

American journal of veterinary research. 56(10):1322-1324.

Tavornpanich, S., W. O. Johnson, R. J. Anderson, and I. A. Gardner. 2008. Herd Characteristics and Management Practices Associated with Seroprevalence of Mycobacterium avium subsp. paratuberculosis Infection in Dairy Herds. American Journal of Veterinary Research 69(7):904-

911.

Thamsborg, S. M., A. Roepstorff and M. Larsen. 1999. Integrated and Biological Control of

Parasites in Organic and Conventional Production Systems. Veterinary Parasitology 84: 169-186.

Thompson-Crispi, K. A., B. Hine, M. Quinton, F. Miglior, and B. A. Mallard. 2012. Short

Communication: Association of Disease Incidence and Adaptive Immune Response in Holstein

Dairy Cows. Journal of Dairy Science 95(7):3888-3893.

Tiwari, A., J. A. VanLeeuwen, I. R. Dohoo, G. P. Keefe, J. P. Haddad, H. M. Scott, and T.

Whiting. 2009. Risk Factors Associated with Mycobacterium avium subspecies paratuberculosis

Seropositivity in Canadian Dairy Cows and Herds. Preventive Veterinary Medicine 88(1):32-41.

Tkachuk, V. L., D. O. Krause, T. A. McAllister, K. E. Buckley, T. Reuter, S. Hendrick, and K. H.

Ominski. 2013. Assessing the Inactivation of Mycobacterium avium subsp. paratuberculosis

During Composting of Livestock Carcasses. Appl Environ Microbiol 79(10):3215-3224. van Hulzen, K. J. E., A. P. Koets, M. Nielen, H. C. M. Heuven, J. A. M. van Arendonk, and D.

Klinkenberg. 2014. The Effect of Genetic Selection for Johne's Disease Resistance in Dairy cattle: Results of a Genetic-Epidemiological Model. Journal of Dairy Science 97(3):1762-1773.

VanLeeuwen, J. A., G. P. Keefe, R. Tremblay, C. Power, and J. J. Wichtel. 2001. Seroprevalence of Infection with Mycobacterium avium subspecies paratuberculosis, Bovine Leukemia Virus,

and Bovine Viral Diarrhea Virus in Maritime Canada Dairy Cattle. Canadian Veterinary Journal

42:193-198.

VanLeeuwen, J. A., A. Tiwari, J. C. Plaizier, and T. L. Whiting. 2006. Seroprevalences of

Antibodies Against Bovine Leukemia Virus, Bovine Viral Diarrhea Virus, Mycobacterium avium subspecies paratuberculosis, and Neospora caninum in Beef and Dairy Cattle in Manitoba.

Canadian Veterinary Journal 47:783-786

VanLeeuwen, J. A., L. Forsythe, A. Tiwari, and R. Chartier. 2005. Seroprevalence of Antibodies

Against Bovine Leukemia Virus, Bovine Viral Diarrhea Virus, Mycobacterium avium subspecies paratuberculosis, and Neospora caninum in Dairy Cattle in Saskatchewan. Can Vet J 46(1):56-

58. van Roermund, H. J. W., D. Bakker, P. T. J. Willemsen, and M. C. M. de Jong. 2007. Horizontal

Transmission of Mycobacterium avium subsp. paratuberculosis in Cattle in an Experimental

Setting: Calves can Transmit the Infection to Other Calves. Veterinary Microbiology 122(3-

4):270-279.

Wagenaar, J. P., P. Klocke, G. Butler, G. Smolders, J. H. Nielsen, A. Canever and C. Leifert.

2011. Effect of Production System, Alternative Treatments and Calf Rearing System on Udder

Health in Organic Dairy Cows. Njas-Wageningen Journal of Life Sciences 58: 157-162.

Weber, M. F., J. Kogut, J. de Bree, G. van Schaik, and M. Nielen. 2010. Age at which Dairy

Cattle Become Mycobacterium avium subsp. paratuberculosis Faecal Culture Positive.

Preventive Veterinary Medicine 97(1):29-36.

Wells, S. J. and B. A. Wagner. 2000. Herd-Level Risk Factors for Infection with Mycobacterium paratuberculosis in US Dairies and Association Between Familiarity of the Herd Manager with the Disease or Prior Diagnosis of the Disease in that Herd and use of Preventive Measures.

JAVMA 219(9):1450-1457.

Whitlock, R. H., S. J. Wells, R. W. Sweeney, and J. Van Tiem. 2000. ELISA and Fecal Culture for Paratuberculosis (Johne's Disease): Sensitivity and Specificity of Each Method. Veterinary

Microbiology 77(3-4):387-398.

Whittington, R. J., D. J. Marshall, P. J. Nicholls, I. B. Marsh, and L. A. Reddacliff. 2004.

Survival and Dormancy of Mycobacterium avium subsp. paratuberculosis in the Environment

Applied and Environmental Microbiology 70(5):2989-3004.

Whittington, R. J., I. B. Marsh, and L. A. Reddacliff. 2005. Survival of Mycobacterium avium subsp. paratuberculosis in Dam Water and Sediment. Appl Environ Microbiol 71(9):5304-5308.

Whittington, R. J. and P. A. Windsor. 2009. In Utero Infection of Cattle with Mycobacterium avium subsp. paratuberculosis: A Critical Review and Meta-Analysis. The Veterinary Journal

179(1):60-69.

Windsor, P. A. and R. J. Whittington. 2010. Evidence for Age Susceptibility of Cattle to Johne's

Disease. The Veterinary Journal 184(1):37-44.

World Organisation for Animal Health (OIE). 2012. Paratuberculosis (Johne's Disease). Pages

276-291 in Manual of Diagnostic tests and Vaccines for Terrestrial Animals. Vol. 1. 7 ed.

Wraight, M. D., J. McNeil, D. S. Beggs, R. K. Greenall, T. B. Humphris, R. J. Irwin, S. P. Jagoe,

A. Jemmeson, W. F. Morgan, P. Brightling, G. A. Anderson, and P. D. Mansell. 2000.

Compliance of Victorian Dairy Farmers with Current Calf Rearing Recommendations for Control of Johne's Disease. Veterinary Microbiology 77(3-4):429-442.

Zwald, A. G., P. L. Ruegg, J. B. Kaneene, L. D. Warnick, S. J. Wells, C. Fossler and L. W.

Halbert. 2004. Management Practices and Reported Antimicrobial Usage on Conventional and

Organic Dairy Farms. Journal of Dairy Science 87: 191-201.

CHAPTER 2

VARIABILITY IN THE RISK ASSESSMENT AND MANAGEMENT

PLAN (RAMP) SCORES COMPLETED AS PART OF THE ONTARIO

JOHNE’S EDUCATION AND MANAGEMENT ASSISTANCE

PROGRAM (2010-2013)

(This chapter has been formatted for submission to The Canadian Veterinary Journal)

Laura Pieper1, Ann Godkin2, Karen Hand3, Nicole Perkins1, Jamie Imada1, Ulrike Sorge4, Trevor

DeVries5, David F Kelton1

1 Department of Population Medicine, University of Guelph, Ontario, Canada, N1G 2W1

2 Veterinary Science and Policy Group, Ontario Ministry of Agriculture and Food/Ministry of

Rural Affairs (OMAF/MRA), Ontario, Canada, N1G 4Y2

3 Strategic Solutions Group, Ontario, Canada, N0B 2J0

4 Department of Veterinary Population Medicine, University of Minnesota, St. Paul, USA, MN

55108

5 Department of Animal and Poultry Science, University of Guelph, Ontario, Canada, N1G 2W1

Corresponding author: David Kelton

[email protected]

2.1 ABSTRACT

As a proactive measure towards controlling the non-treatable and contagious Johne’s disease in cattle, the Ontario dairy industry launched the Ontario Johne’s Education and Management

Assistance Program in 2010. The objective of this study was to describe the results of the first four years of the program and to investigate the variability in the Risk Assessment and

Management Plan (RAMP) scores associated with the county, veterinary clinic and veterinarian.

Of 4,158 Ontario dairy farms, 2,153 (51.8 %) participated in the program between January 2010 and October 2014. For this study, RAMP scores and whole herd milk or serum ELISA results were available from 2,103 farms. Herd-level ELISA positive prevalence (herds with one or more test positive cows were considered positive) was 27.2%. Linear mixed model analysis revealed that the greatest RAMP score variability was at the veterinarian level (24.2%), with relatively little variability at the county and veterinary clinic levels. Consequently, annual RAMPs should be done by the same veterinarian to avoid misleading or discouraging results.

2.2 INTRODUCTION

Johne’s disease (JD) or paratuberculosis is a chronic infectious disease of ruminants associated with granulomatous enteritis caused by Mycobacterium avium spp. paratuberculosis (MAP).

Estimates of the apparent JD herd-level seroprevalence in dairy herds among Canadian provinces vary between 33 to 74% (Tiwari et al., 2006).

Diagnostic and screening tests usually have a low sensitivity compared to tissue culture or faecal culture as the gold standard, especially when the prevalence in the herd is low or the individual is shedding few bacteria (Tiwari et al., 2006). Therefore, test-and-cull programs alone are considered less effective than risk-based prevention strategies for JD control (Kudahl et al.,

2008). Risk prevention strategies aim to reduce the risk of disease transmission through improvements in targeted management practices.

The voluntary Ontario Johne’s Education and Management Assistance Program (OJEMAP) was launched in 2010 (www.johnes.ca). A standardized management questionnaire, the Risk

Assessment and Management Plan or RAMP, was administered by trained, private veterinarians.

Testing of milk or serum from all lactating cows on a single date was used to identify MAP antibody ELISA positive cows. For herds participating in milk recording through CanWest Dairy

Herd Improvement (DHI), milk was collected from each cow through milk recording meters and submitted to the DHI laboratory for automated analysis. The PARACHEK milk ELISA (Prionics

AG, Schlieren, Switzerland) used in the program was found to have a sensitivity of 61.1% and a specificity of 94.7% compared to fecal culture (Hendrick et al., 2005a). Cows were considered negative at a corrected optical density (OD) <0.00, suspicious at OD 0.00-0.09, positive at OD

0.1-0.99, or high positive (HP) at OD ≥1.0. For herds not participating in milk recording or for additional dry cow testing, serum samples taken by the herd veterinarian were submitted to the

Animal Health Laboratory (University of Guelph, Ontario, Canada) for analyses using the

IDEXX HerdChek Mycobacterium paratuberculosis ELISA (IDEXX Laboratories, Westbrook,

Maine, USA). Cows were considered negative at a sample-to-positive control ratio (S:P) <0.10, suspicious at S:P 0.10-0.24, positive at S:P 0.25-0.99, or HP at S:P ≥1.0, and the sensitivity and specificity compared to fecal culture were 73.6% and 87.5%, respectively (Hendrick et al.,

2005a).

The RAMP was a semi-quantitative tool, with low scores indicating a low risk of disease transmission. It contained five risk assessment sections including: 1) general JD management and biosecurity (maximum 60 points), 2) calving area risk management (maximum 80 points), 3) pre- weaned heifer management (maximum 70 points), 4) post-weaned heifer management (maximum

40 points), and 5) adult cow and manure management (maximum 50 points). The maximum overall RAMP score combining all 5 sections was 300 points. A copy of the RAMP is found in

Appendix I.I. Farm specific recommendations for management changes to reduce MAP transmission in high-risk areas were made by the veterinarian. When cows with HP ELISA test results were identified, farmers were encouraged to dispose of these individuals within 90 days of testing while ensuring that these animals were not moved to another dairy herd or into the human food chain. Expenses for the ELISA testing ($8 CAD/cow) and cow removal compensation ($500

CAD/disposed high titre cow) were covered by the program for every farm participating for the first time, if high positive cows were removed in time. In 2013, all farms were offered a second program funded test. The expenses for the assessing veterinarian to administer the RAMP were covered by the farmers. Full implementation of the program was completed in August 2013; however, a formal description of the results has not yet been reported.

The RAMP used in the OJEMAP was adapted from the risk assessment used by Sorge et al.

(2011) and is based on identified high risk areas for MAP transmission. The authors showed that farms participating in the risk assessment-based control program could reduce their risk scores as well as the apparent within-herd milk ELISA prevalence for JD. Through the OJEMAP, more than 95% of Ontario dairy veterinarians have been trained and registered to conduct the risk assessment for their clients. A RAMP manual (Appendix I.II) was provided to veterinarians for further consistency of assessments. Although there have been attempts to standardize the RAMP currently in use, many parts include subjective measurements and observations that could be influenced by the perspective and biases of the assessing veterinarian. Additionally, specific veterinary clinic policies or regional differences in the diversity of dairy herds might introduce variability at those levels that are not attributable to the farm situation itself.

Therefore, the objectives of this study were to: 1) describe the participation, RAMP scores and

ELISA prevalence in the OJEMAP, and 2) assess the variability of RAMP scores associated with the county, veterinary clinic, veterinarian and herd.

2.3 MATERIALS AND METHODS

2.3.1 Data The dataset was extracted from the OJEMAP database and included RAMPs completed between

January 2010 and August 2013 and the corresponding herd Johne’s ELISA test results. All but the first duplicate observations (since some herds had tested more than once) and herds that included only a RAMP score or an ELISA test result, but not both, were removed. A flowchart of the data management is displayed in Figure 2.1. Information about county in which the herd was located, the name of the veterinarian who completed the RAMP and their veterinary clinic, number of cows tested with either milk or serum ELISA, number of ELISA-positive cows, number of ELISA HP cows, the HP disposal method, the overall RAMP score and section scores, and the recommendations given to the producers were available for each farm. A farm was considered JD Herd ELISA Status (HES) positive, if they had at least one cow with a JD ELISA positive or high positive result. JD ELISA suspect cows were considered negative for this analysis.

The recommendations that were given to the farmers after the RAMP were coded with numerical codes by a single person (JI). For this study, the recommendations were manually associated with the respective RAMP questions and sections in the RAMP based on reference of recommendations to risk described and assessed in this area.

2.3.2 Statistical analyses The RAMP section scores were calculated by adding up the respective question scores in each section. Normality of the data was assessed visually using histograms. For the descriptive statistics, the geometric mean was reported for non-normally distributed data (e. g. section scores). A small number (1) was added to variables including null values to allow for proper transformation.

There was a non-linear relationship between overall RAMP score and number of test positive cows. Consequently, the models were built with the dichotomous variable HES (1=one or more

ELISA positive animals; 0= zero ELISA positive animals). This variable was included in the model to account for differences in JD ELISA prevalence among veterinarians, veterinary clinics or regions.

A hierarchical approach to model building was started with a linear regression model with RAMP score as the outcome and HES as the fixed effect. Each level of the hierarchy (i.e. veterinarian, veterinary clinic, county) was first tested alone as a random effect for explanation of RAMP score variability. Then, starting at the veterinarian level, higher levels (veterinary clinic and county) were offered to the previous model as random effects. A random slope was tested to determine whether the strength of the relationship between HES and RAMP score was dependent on the assessing veterinarian. Bayesian Information Criteria (BIC) and likelihood ratio tests were used to find the best model. The model fit was assessed graphically at each hierarchical level by plotting the residuals against the predicted value and the inverse normal score.

The statistical data analyses were conducted using STATA 10.1 (StataCorp, Texas, USA). A probability level of P < 0.05 was considered significant. Geographical maps displaying the number of herds per county and the proportion of herds participating per county were generated using ArcMap 10 (ESRI, Redlands, California, USA).

2.4 RESULTS

Fifty two percent of all herds over all counties in Ontario participated in the program, but the proportion of participating herds per county varied between 8.8 to 86.1%. Areas with high participation were not necessarily areas with high herd density (Figure 2.2). After data cleaning,

2,103 herd observations with complete RAMP information and ELISA results were retained in the dataset. There was considerable variation in program participation at the county, veterinary

clinic, and veterinarian level (Table 2.1). Some veterinarians conducted up to 63 RAMPs whereas others conducted only one RAMP. Of all tested farms, 572 (27.2%) had at least one, 261 (12.4%) had at least two JD ELISA positive cows, and 130 (6.2%) had at least one HP cow in their herd.

Of 106 farms where the HP disposal method was noted, 19 (18%) buried, 24 (23%) composted, and 63 (59%) used a deadstock service to dispose their HP animals. The average number of JD

ELISA positive cows and the average within-herd ELISA prevalence was 0.34 (95% CI: 0.31-

0.37) and 0.85% (95% CI: 0.76-0.94%) for all tested herds, and 1.93 (95% CI: 1.80-2.07) and

3.16% (95% CI: 2.93-3.40%) for HES positive herds, respectively. The average overall RAMP score was 121.3 (95% CI: 119.7-122.8) on all and 129.8 (95% CI: 126.8-132.8) on HES positive farms. Section 1 score was clearly bimodal with many farmers having a zero value, indicating that the risk of introducing JD to the farm was very low. Therefore, neither an arithmetic mean nor a geometric mean as a measure of central tendency describes the data appropriately. The proportion of herds with a section 1 score of zero was 27.5% in all herds and 20.8% in HES positive herds. Detailed descriptions of RAMP scores, number of cows tested, and within herd

ELISA prevalence for all herds and JD HES positive herds are displayed in Table 2.2.

After completing the initial risk assessment, the veterinarians gave zero (0.5%), one (10.0%), two

(24.6%), or three (64.9%) recommendations to the farmers (mean: 2.54, median: 3) for changes to improve JD control on their farm. The most common recommendation, with 44.1% of participating farmers receiving it, was to not purchase cows, to minimize purchases, or to buy only from known low-risk herds. Table 2.3 displays the 14 most common recommendations that were each given to over 5% of herds.

When tested as the only random effect, 5%, 19%, and 24% of the RAMP total score was explained by the county, the veterinary clinic and the veterinarian, respectively. Any additional variation in the RAMP score explained at the county, or veterinary clinic level was not significant

when the veterinarian was in the model. The random slope did not contribute to the explanation of variation in the model. Confounding by veterinarian was not detected. On HES positive farms the RAMP score was 11.7 points higher than on negative farms (Table 2.4). In the final model, the likelihood ratio test (against a linear model without random effects) was highly significant

(p<0.001) and residuals at both levels were normally distributed and homoscedastic, indicating good model fit.

2.5 DISCUSSION

This paper describes results from the voluntary Johne’s disease control program in Ontario,

Canada. Of 2,103 herds included in this analysis, 27.2% had at least one JD milk ELISA positive animal. This estimate is lower than the 34% (95% CI 21-47%) milk ELISA prevalence estimate reported by Hendrick et al. (2005b). However, the confidence intervals in the two studies are quite wide and do overlap, indicating that the estimates are not significantly different.

The most common recommendation made by veterinarians to dairy farmers (44%) was to keep their herd closed or to be more cautious when purchasing animals. This may indicate that veterinarians perceive that producers potentially have insufficient screening and isolation practices when purchasing animals and they should therefore avoid bringing cattle in, especially if they are low prevalence herds. The introduction of cattle into the herd has been identified as a risk factor for JD herd positive results in other studies (Sorge et al., 2012; Wells and Wagner,

2000). In this study, 72.5% of producers had a non-zero score in section 1, showing that they had some form of herd additions within the last 5 years. On the other hand, there were 21% HES positive producers with a zero score in section 1, indicating that they had not purchased animals.

However, this measure does not account for earlier herd additions and herds may have introduced infectious cattle more than 5 years ago. Consequently, it might be that Ontario dairy farmers were

benefitting from a larger educational campaign to increase awareness regarding risks of disease introduction when purchasing animals (“Buyer Beware”).

The RAMP score was 12 points higher on HES positive farms compared to HES negative farms.

One question in the risk assessment specifically refers to the HES (Section 2 Question 5: Calving area used by JD clinical or test positive cows?) with the answer categories 1, 4, 7, or 10 based on increasing JD prevalence. However, this question alone could not account for the observed

RAMP score difference between HES positive and negative herds. This suggests that there is a positive association between positive HES and RAMP score beyond this specific question.

The county-level ICC in our study was very low at 5%, which is similar to the estimate by

Berghaus et al. (2005) of the state-level correlation in the United States. Furthermore, variation at the veterinary clinic level was explained by the respective individual veterinarians. This indicates that there is no similarity in RAMP scores among veterinarians from the same clinic. The variation at the veterinarian level was about 24%, indicating that RAMPs conducted by one veterinarian are more similar to each other than to those from other veterinarians. This estimate is higher than the estimate from Berghaus et al. (2005), who reported 15% variation from the

National Animal Health Monitoring System (NAHMS) Dairy 2002 study. However, the NAHMS data collection was completed within a 7 month period, and due to the random selection of farms, it is likely that the assessors (federal or state veterinary officers, or animal health technicians) did not have an established relationship with the farms. One might hypothesize for our population that herds served by one veterinarian are more similar to each other as compared to herds visited by other veterinarians, due to veterinary-client relationship that might foster unique and consistent advice to each veterinarian’s client herds. It could also be that one veterinarian gave generally higher or lower scores for a specific area than other veterinarians based on their perceptions, beliefs and biases. However, only the random intercept but not random slope was

significant. That indicates that the association between RAMP score and HES was consistent, regardless of the assessing veterinarian. Nevertheless, for herd comparisons or repeated assessments, the detected variability in the RAMP scores might cause misleading results in cases where the farmer did not implement changes but, because the assessor changed, a positive change is documented in the RAMP. Conversely, results might be disappointing or discouraging if the producer implemented changes but, because the assessing veterinarian changed, the efforts were not recognized by a lower RAMP score. Consequently, when assessments of management improvement are done repeatedly and are used to document improvement or change, all assessments for a given farm should be done by the same veterinarian to avoid misleading results.

There was a high degree of variability of the intensity of program participation among counties, veterinary clinics and veterinarians. Some veterinarians assessed considerably more farms than others. It might be that very active veterinarians perceive Johne’s disease as a greater problem among their clients than others. It could also be that producers in those areas with low participation were less aware of the program and did not ask their veterinarians for this service.

Moreover, there were anecdotes about DHI technicians actively enrolling clients into the program in some regions which might have additionally increased participation and caused differences among counties. Nevertheless, education for Johne’s disease prevention and advertisement for participation in a control program should be intensified at the different levels.

2.6 CONCLUSION

The RAMP scores were higher for JD ELISA positive herds than for negative herds. There was considerable variation in RAMP scores at the level of the veterinarian, but not at the practice or county level. RAMPs from the same veterinarian were more similar to each other than to those from other veterinarians and consecutive assessments in one farm should be done by the same veterinarian. Regional participation rates and the different involvement of veterinarians in the

program indicate that educational advertisement on different levels would be necessary to improve overall program participation.

2.7 REFERENCES

Berghaus, R. D., J. E. Lombard, I. A. Gardner, and T. B. Farver. 2005. Factor Analysis of a

Johne's Disease Risk Assessment Questionnaire with Evaluation of Factor Scores and a Subset of

Original Questions as Predictors of Observed Clinical Paratuberculosis. Preventive Veterinary

Medicine 72(3-4):291-309.

Hendrick, S., T. Duffield, D. Kelton, K. Leslie, K. Lissemore, and M. Archambault. 2005a.

Evaluation of Enzyme-Linked Immunosorbent Assays Performed on Milk and Serum Samples for Detection of Paratuberculosis in Lactating Dairy Cows. Journal of the American Veterinary

Medical Association 226(3):424-428.

Hendrick, S., T. Duffield, K. Leslie, K. Lissemore, M. Archambault, and D. Kelton. 2005b. The

Prevalence of Milk and Serum Antibodies to Mycobacterium avium subspecies paratuberculosis in Dairy Herds in Ontario. Can Vet J 46(12):1126-1129.

Kudahl, A. B., S. S. Nielsen, and S. Østergaard. 2008. Economy, Efficacy, and Feasibility of a

Risk-Based Control Program against Paratuberculosis. Journal of Dairy Science 91(12):4599-

4609.

Sorge, U. S., K. Lissemore, A. Godkin, J. Jansen, S. Hendrick, S. Wells, and D. F. Kelton. 2011.

Changes in Management Practices and Apparent Prevalence on Canadian Dairy Farms

Participating in a Voluntary Risk Assessment-Based Johne's Disease Control Program. Journal of

Dairy Science 94(10):5227-5237.

Sorge, U. S., K. Lissemore, A. Godkin, J. Jansen, S. Hendrick, S. Wells, and D. F. Kelton. 2012.

Risk Factors for Herds to Test Positive for Mycobacterium avium ssp. paratuberculosis-

Antibodies with a Commercial Milk Enzyme-Linked Immunosorbent Assay (ELISA) in Ontario and Western Canada. Can Vet J 53(9):963-970.

Tiwari, A., J. A. VanLeeuwen, S. L. B. McKenna, G. P. Keefe, and H. W. Barkema. 2006.

Johne's Disease in Canada Part I: Clinical Symptoms, Pathophysiology, Diagnosis, and

Prevalence in Dairy Herds. Canadian Veterinary Journal 47:874-882.

JAVMA 219(9):1450-1457.

2.8 TABLES

Table 2.1 Distribution of Risk Assessment and Management Plans (RAMPs) at the county, veterinary clinic and veterinarian level Number of Mean number of 95% Confidence Minimum Maximum categories RAMPs1 Interval County 47 24.7 17.2-35.5 1 187 Veterinary clinic 107 9.1 7.0-11.7 1 122 Veterinarian 184 7.7 6.7-8.9 1 63 1 Geometric mean

Table 2.2 Description of ELISA test results and Risk Assessment and Management Plan (RAMP) scores in all herds and Herd ELISA Status (HES) positive herds participating in the Ontario Johne’s Education and Management Assistance Program All Herds (n=2103) HES Positive Herds (n=572) Mean1 95% CI2 Minimum Maximum Mean1 95% CI Minimum Maximum Number of cows tested 54.1 52.9-55.4 12 986 74.2 70.3-78.4 18 986 RAMP score3, 4 121.3 119.7-122.8 26 231 129.8 126.8-132.8 28 220 Section 1 score5 14.9 13.7-16.1 0 60 20.2 17.6-23.1 0 60 Section 2 score6 30.8 30.3-31.4 8 74 33.2 32.2-34.3 8 74 Section 3 score7 19.4 19.0-19.9 7 58 19.9 19.1-20.8 7 58 Section 4 score8 14.6 14.2-14.9 4 40 13.9 13.3-14.6 4 40 Section 5 score9 13.7 13.4-14.1 4 45 15.5 14.7-16.3 4 44 Number ELISA positive cows 0.34 0.31-0.37 0 33 1.93 1.80-2.07 1 33 Within herd prevalence10 of ELISA 0.85 0.76-0.94 0 22.6 3.16 2.93-3.40 0.3 22.6 positive cows [%] Number of ELISA high positive cows 0.052 0.043-0.062 0 5 0.21 0.17-0.24 0 5 Within herd prevalence10 of ELISA 0.11 0.08-0.13 0 8.7 0.40 0.32-0.48 0 8.7 high positive cows [%] 1 Geometric mean if not indicated otherwise 2 95% CI = 95% Confidence Interval 3 Maximum possible score is 300 4 Arithmetic mean displayed 5 Section 1 = General Johne’s Disease Risk Management (maximum = 60 points) 6 Section 2 = Calving Area Risk Management (maximum = 80 points) 7 Section 3 = Heifers – Pre-weaned Risk Management (maximum = 70 points) 8 Section 4 = Heifers – Weaned to First Calving Risk Management (maximum = 40 points) 9 Section 5 = Cows – Risk Management (maximum = 50 points) 10 Calculated as the number of (high) positive animals /number of cows tested in a herd

Table 2.3 Most frequent recommendations for Johne’s disease prevention and control given by veterinary assessors to producers after the Risk Assessment and Management Plan (RAMP) Recommendation Number of farms this Attributable RAMP recommendation was section2 given to (proportion [%]1) Don’t purchase more cows / minimize purchases / buy 927 (44.1) 1 from low-risk herds you know Remove heifer calves ASAP/quickly (when standing) 690 (32.8) 2 from maternity area / pasture to individual pens / hutches Feed more colostrum and on time 338 (16.1) 3 Don’t feed non-saleable milk or high SCC milk to 331 (15.7) 3 heifer calves Separate newborn calf from cow – create mini-pen, 309 (14.7) 2 calf box/cart, Rubbermaid tub One cow in calving pen at a time/ minimize cows in 202 (9.6) 2 calving pen Hospital pen is not the calving pen (need separate pen) 197 (9.4) 2 Separate heifers (bred/pregnant/breeding) from dry 166 (7.9) 4 cows Retest herd in 12-18 months/ continue testing 157 (7.5) 1, 2 Feed low-risk milk 142 (6.8) 3 Add more bedding/ reduce stocking density/ cleaner 140 (6.7) 4 heifers Create additional maternity pen (in dry cow area / 126 (6.0) 2 main barn) or create dedicated maternity pen in barn Move calving cows to individual pen from group area 125 (5.9) 2 / pasture/yard / tiestalls (don’t calve in group area, separate her from group) Feed milk replacer / minimize whole milk feeding 116 (5.5) 3 especially prior to going on replacer 1 Percentages do not add up to 100 because farmers could be given up to three recommendation 2 See Table 2.2 for explanation of RAMP sections

Table 2.4 Final mixed linear regression model predicting the overall RAMP score in the herd. Herds cluster by assessing veterinarian (Intra-class Correlation Coefficient=0.24) Coefficient SE p-Level Intercept 117.86 1.626 <0.001 HES1 Positive Farm 11.72 1.599 <0.001 Variance SE Veterinarian (n=184) 308.90 45.24 Herd (n = 2103) 967.13 31.15 1 HES=Herd ELISA Status

2.9 FIGURES

Figure 2.1 Flowchart of number of observations at each stage from combining the RAMP data with the ELISA data until the final number of observations analyzed

Figure 2.2 Choropleth maps of number of herds per county and proportion of herds per county participating in the Ontario Johne’s Education and Management Assistance Program in southern Ontario

CHAPTER 3

EVALUATION OF THE JOHNE’S DISEASE RISK ASSESSMENT AND

MANAGEMENT PLAN IN DAIRY HERDS IN ONTARIO, CANADA

Laura Pieper1, Ulrike Sorge2, Ann Godkin3, Trevor DeVries4, Kerry Lissemore1, David Kelton1

1Department of Population Medicine, University of Guelph, Ontario, Canada, N1G 2W1

2Department of Veterinary Population Medicine, University of Minnesota, St. Paul, USA, MN

55108

3Veterinary Science and Policy Group, Ontario Ministry of Agriculture and Food (OMAF),

Ontario, Canada, NOB 1S0

4Department of Animal and Poultry Science, University of Guelph, Ontario, Canada, N1G 2W1

Corresponding author: David Kelton; [email protected]

3.1 ABSTRACT

Johne’s disease (JD) is a production limiting, gastrointestinal disease in cattle. To minimize the effects of JD, the Ontario dairy industry launched the Ontario Johne’s Education and

Management Assistance Program in 2010. As part of the program, trained veterinarians conducted a Risk Assessment and Management Plan (RAMP), an on-farm questionnaire, where high RAMP scores are associated with high risk of JD transmission. Subsequently, veterinarians recommended farm specific management practices for JD prevention. Milk or serum ELISA results from the milking herd were used to determine the Herd ELISA Status (HES). After three and a half years of implementation of the program, the aim of this study was to evaluate the associations among RAMP scores, HES, and recommendations. Data from 2,103 herds were available for the analyses. Mixed logistic regression models showed a positive association

between HES and general JD risk, calving area risk, pre-weaned calf area risk, adult cow area risk, as well as the overall RAMP score. Generally, there was a positive association between

RAMP scores and the odds of receiving a recommendation for the respective risk area; however, the relationship was not always linear and an interaction between HES and RAMP section score for general JD risk and calving area risk was observed. Two separate zero-inflated negative binomial models for the prediction of the number of ELISA positive animals per farm were built.

Both models included questions about purchasing animals in the logistic portion, indicating risks for between-herd transmission, and purchasing bulls, colostrum and milk feeding management, and adult cow environmental hygiene in the negative binomial portion, indicating risk factors for within-herd transmission. However, the count ratio was below 1 if farms fed low risk milk compared to milk replacer, indicating a preventative effect of the former practice. Model A additionally included the JD herd history in the negative binomial and the logistic portion as well as the birth of calves outside the designated calving area in the negative binomial potion. In

Model B, there was a negative association between the frequency of feeding non-saleable milk to calves and the number of ELISA positive animals on the farm.

This study suggests that the RAMP is a valuable tool to assess the risk for JD transmission within and between herds and to determine farm specific recommendations for JD prevention.

3.2 INTRODUCTION

In many countries worldwide, Johne’s disease (JD) has been acknowledged as a production limiting, infectious, gastrointestinal disease in dairy cattle. The disease is caused by

Mycobacterium avium spp. paratuberculosis (MAP), a very slow growing and intracellular bacterium with a wide host spectrum. In Canada, economic losses due to JD have been estimated

at $1,196 CAD per 100 cows for the entire dairy industry, and $2,992 CAD in an average-sized, seropositive dairy herd (Tiwari et al., 2008). Losses are attributable to a drop in milk yield, reduced slaughter weight and early culling. Furthermore, JD has been associated with Crohn’s disease, a chronic gastrointestinal disease in humans. However, evidence supporting a zoonotic potential for MAP is not very strong and a definitive link between JD and Crohn’s disease has yet to be established (Waddell et al., 2008).

To limit the effects of JD on the cattle industry many countries, including Canada, have implemented voluntary control and surveillance programs. Contrary to many other diseases in which control programs focus on testing and culling of infected animals, JD control is most commonly based on identification and reduction of transmission risk. MAP infection is not easily detected clinically or with available diagnostic tests and infected animals can remain undetected for years. While calves have the highest probability of getting infected (Windsor and

Whittington, 2010), the highest probability of being detected is in older animals. Detection by fecal culture is greatest at an age between 2.5 and 5.5 years, and by antibody titers at an age between 2.5 and 4.5 years (Nielsen and Ersbøll, 2006). When compared to fecal culture, the sensitivity of JD ELISA in serum or milk has been estimated to be 73.6% and 61.1%, respectively (Hendrick et al., 2005). Therefore, while a traditional test-and-cull program is unlikely to be successful, the tests can help to identify some infected and infectious animals in the herd. A major focus of JD control programs is on preventing the spread of the disease in the early life of calves through best management practices, many of which are aimed at minimizing the exposure of newborn calves to infectious cows as well as their colostrum, milk and manure.

A modeling study by Kudahl et al. (2008) demonstrated that while the prevalence of JD would increase when using a test and cull program only, it would decrease when combining the test and

cull program with improved management over the first 6 months of a calf’s life. Furthermore, an empirical study from Canada showed that farms could reduce their JD prevalence through participation in a formal risk assessment based JD control program (Sorge et al., 2011).

In Ontario, Canada, the Ontario Johne’s Education and Management Assistant Program

(OJEMAP) was piloted in 2006 and launched in January 2010. The program is explained in greater detail on the program website, www.johnes.ca, and in Pieper et al. (CHAPTER 2).

Briefly, all Ontario dairy farms were encouraged to do a Risk Assessment and Management Plan

(RAMP) with their herd veterinarian. To administer the RAMP, trained and registered local veterinarians visited the farm and conducted a risk assessment using a standardized one-page questionnaire. A detailed guide for the RAMP was provided to ensure objectivity of the assessment. The RAMP was adapted from the risk assessment used by Sorge et al. (2011), but the considerably shortened version used in the program had not yet been formally evaluated since the program was launched.

The RAMP consisted of 5 sections focusing on: 1) general JD risk, 2) calving area risk, 3) pre- weaned heifer risk, 4) post-weaned heifer risk, and 5) adult cow risk. Each section was assigned a score, with high risk scores indicating a high risk of disease transmission. Based on information acquired while administering this questionnaire, veterinarians provided up to three farm specific recommendations for improved JD prevention and control.

Each dairy farm in Ontario was eligible to test all milking cows in the herd for Johne’s disease via milk or blood ELISA. The costs of the herd test were covered by the program, if certain requirements were met. The two requirements to be met were that: 1) a RAMP had to be conducted within 90 days of testing and 2) cows that had a high positive test result (milk OD

≥1.0 or serum S:P ≥1.0) were disposed of within 90 days after the test or before the next calving,

whichever came first, while ensuring that the animal did not enter the human food chain and was not sold to another dairy farm.

The objective of this study was to describe the associations among RAMP questions, sections, and overall scores and the herd ELISA results and the recommendations, respectively, using all available data from the first 3.5 years of the OJEMAP.

3.3 MATERIALS AND METHODS

The OJEMAP data base, containing observations from 2,103 herds evaluated between January

2010 and August 2013 was utilized for the analysis. Serum or milk ELISA results from the milking herd, individual question, section and the overall RAMP scores as well as JD control recommendations for each herd were available. The Herd ELISA Status (HES) was considered positive if at least one cow tested positive on the ELISA and was considered negative if all tested animals were ELISA negative. For further data handling procedures and descriptive statistics of the data see Pieper et al. (CHAPTER 2).

The management recommendations that were given to the farmers after the RAMP had been completed were assigned a numerical code by a single person and stored in the program data base. The recommendations were manually associated with the respective questions and sections of the RAMP, based on where the management change would be expected to influence the risk of JD transmission. For the purposes of the data analysis, only the presence (≥ 1 recommendation) or absence (no recommendation) of a recommendation for each question

(QREC) or section (SREC) was considered; additional recommendations for one herd that addressed a single question or section were ignored.

All statistical analyses were conducted using the computer program Stata/IC 13.0 for Windows

(StataCorp LP, Texas, USA). A probability of <0.05 was considered significant.

As shown by Pieper et al. (CHAPTER 2), there was significant clustering of the RAMP scores at the level of the veterinarian, suggesting that veterinarians had specific management areas that they preferred to focus on. Therefore, clustering by veterinarian was accounted for in the statistical analysis by using mixed logistic regression models and by clustered sandwich estimates for the zero-inflated negative binomial model (ZINB). For graphical display of models addressing SREC, the same fixed effects models that were built using mixed logistic regression were built again using simple logistic regression with clustered sandwich estimates.

Univariable and multivariable mixed logistic regression models with either HES or SREC as the outcome variable and RAMP section scores as explanatory variables were built with veterinarian as a random effect. Univariable and multivariable mixed logistic regression models with either

HES or receiving a QREC as the outcome variables and RAMP question scores as explanatory variables are found in Appendix II. The linearity of the relationships was assessed by locally weighted scatterplot smoothing (lowess) and quadratic terms were offered, or data transformations or categorization performed, if necessary. The number of tested animals per farm was log transformed (lntotnotest) and offered to the HES models as a confounding variable.

HES was tested as a confounding variable for the relationship between RAMP scores and the presence of SREC. Confounding variables were only retained if the change in the β-coefficient of the RAMP section score was >30%, regardless if the confounding variable was significant in the model or not. The intra-class correlation coefficient (ICC) was calculated by the latent variable technique (Dohoo et al., 2009; Formula 1).

For the number of ELISA positive animals per farm as the outcome, a ZINB model was chosen based on the assumption that a zero count in a herd could arise from two different processes: 1) a herd has management practices that prevent it from introducing the infection into the herd and 2) there are risk factors (management practices) that could reduce the prevalence within a herd so that the diagnostic tests are unable to detect infected animals at the time of sampling. The inflated portion of the model is interpreted as the odds of having a zero count whereas the negative binomial portion is interpreted as a Count Ratio (CR; Tiwari et al., 2009). A negative coefficient in the inflated portion would therefore indicate an increased odds of having a non- zero count or, in other words, of being a HES positive farm. In the negative binomial portion, a

CR >1 indicates an increase in the number of positive animals on the farm proportional to herds that do not have that risk factor. The first model (Model A) was built by manual forward selection using p<0.05 for the RAMP questions as categorical variables and the lntotnotest as a linear predictor variable if it had a p<0.1 at the univariable level. The variable lntotnotest was included as an offset in the model. As an approximation for herd size but also for herd density, lntotnotest was additionally offered to the model in the logistic and negative binomial portion as a confounder for herd management variables. Herd size may be an approximation for herd density based on the hypothesis by Godkin (2011) where, in a bigger herd, a greater number of calves could be exposed to a potentially infectious animal at the time of birth than in a smaller herd. In a smaller herd, fewer cows are calving at a particular time, which minimizes the exposure of a calf to adult cows other than their own dam, even in herds using a group calving pen. The inflated portion of the model was modelled considering only section 1 variables, which consisted of measures of herd biosecurity and introducing JD into the herd, in addition to lntotnotest. For the negative binomial portion of the model, all variables from every section of

the RAMP were considered. Different models were compared using the Akaike Information

Criteria (AIC) and models with lower AIC were considered superior. The residuals of the final model were calculated by subtracting the predicted from the observed number of positive animals per farm, and were graphically assessed to identify potential outliers. The Vuong test and likelihood ratio test were used to determine superiority of the model compared to a regular negative binomial model or a zero-inflated poisson model, respectively (Dohoo et al., 2009).

Because of redundancy of the variable Section (SEC) 1 Question (QN) 2 (Have you ever had any Johne’s disease positive cows on your farm?), a second ZINB model (Model B) was built using a similar approach as for Model A; but, excluding variable SEC1QN2 as a possible predictor variable. The HES status was likely to be positive if JD clinical or test positive animals had previously been observed on the farm. Therefore, Model B gave more emphasis to the herd management variables for predicting the number of positive animals on the farm, assuming that

JD was present at some level. Only biologically plausible interactions were tested, and if significant, the main effects were retained in the final model, even if their associated p-values were >0.05. For both ZINB models, the variable SEC2QN5 (Calving of JD positive cows in the calving area?) was not offered to the model because the scores were given based on the known within-herd ELISA positive prevalence of the herd and the cause-effect relationship is reverse for this variable. The predicted and observed probabilities of ELISA negative or number of

ELISA test positive cows were calculated and displayed for Models A and B (Long and Freese,

2006).

3.4 RESULTS

In the univariable and multivariable mixed logistic regression models, positive relationships were found between HES and the overall RAMP score as well as most of the SEC scores, with the

exception of SEC 4 score (i.e. weaned heifer management). The relationships between HES and

SEC scores 3, 4, and 5 were confounded by lntotnotest, the log total number of animals tested.

The variability in HES explained by the veterinarian was low, ranging between 4.0 and 7.7%

(Table 3.1).

Table 3.1 Mixed logistic regression analyses investigating the association between herd ELISA status and Risk Assessment and Management Plan (RAMP) section scores and overall RAMP score (controlled for veterinarian as a random effect) RAMP Score OR1 SE2 p-value ICC3 Section 1 [categorical (0/1)]4 1.65 0.20 <0.001 0.067 Section 25 1.03 0.005 <0.001 0.072 Section 36 1.017 0.006 0.002 0.044 Ln(total number tested) 4.54 0.49 <0.001 Section 47 1.009 0.0072 0.209 0.040 Ln(total number tested) 4.46 0.48 <0.001 Section 58 1.025 0.0067 <0.001 0.042 Ln(total number tested) 4.16 0.44 <0.001 Overall 1.011 0.0016 <0.001 0.077 1 OR= Odds Ratio 2 SE= Standard Error 3 ICC= intra class correlation coefficient; shows the amount of clustering by veterinarian 4 Section 1 = General Johne’s Disease Risk Management (Range: 0-60 points) 5 Section 2 = Calving Area Risk Management (Range: 8-74 points) 6 Section 3 = Heifers – Pre-weaned Risk Management (Range: 7-58 points) 7 Section 4 = Heifers – Weaned to First Calving Risk Management (Range: 4-40 points) 8 Section 5 = Cows – Risk Management (Range: 4-45 points)

Table 3.2 and Figure 3.1 demonstrate the relationships between SREC and the RAMP SEC score for that area. Generally, there was a positive association between SEC scores and SREC; however, the relationship was not linear for SEC 3 and 5. For SEC 3, although the model predicts a decreasing probability of receiving a SREC after a score peak at 37 points, the model prediction might not be very precise in that area. Due to the sparse number of observations in that area, the confidence intervals get wider and so the upper bound continues to predict a slowly increasing probability. The probability of receiving a SREC for SEC 1, 2, and 5 was influenced by HES. An interaction was present for the relationship between receiving a recommendation for

general JD risk management and SEC 1 score, as well as receiving a recommendation for calving area risk management and SEC 2 score. In both cases, farms with a low section score and a negative HES were less likely to receive a recommendation compared to farms with a low section score and a positive HES. The probability of receiving a recommendation was not different for HES positive or negative farms if they received high section 1 and 2 scores. For section 5, HES negatively influenced the probability of receiving a recommendation, indicating that producers would be less likely to receive recommendations if their HES was positive. It is noteworthy that the level of the predicted probability of receiving a recommendation varied substantially, from approaching 100% for SEC 2 (Panel b, Figure 3.1) to <24% for SEC 5 (Panel e, Figure 3.1). Concurrently, between 15% (SEC 3) up to 39% (SEC 1) of the variability of receiving a recommendation was explained by the assessing veterinarian (Table 3.2).

Figure 3.2 shows the observed and predicted probability of the number of ELISA positive animals per farm (upper panel) and the difference between observed and predicted probability of the number of ELISA positive animals per farm (lower panel) based on the ZINB models. The predicted and observed probabilities (Figure 3.2, upper panel) are overlapping, indicating good model predictions for both models. The models slightly overestimate the probability of zero counts, underestimate the probability of farms having one test positive cow, and again overestimate the probability of farms having two to seven ELISA positive animals (Figure 3.2, lower panel). Overall, it appears that the amplitude of those over and underestimations is less severe with Model B than Model A. The Vuong tests (Model A z=4.80, p<0.001; Model B z=3.42, p<0.001) and the likelihood ratio tests (Model A Χ2=284.5, p<0.001; Model B

Χ2=325.8, p<0.001) indicated the superiority of the ZINB models compared to regular negative binomial models or zero-inflated poisson models, respectively. The two models share some of

the risk factor variables whereas others occur in one model only (e.g. SEC2QN6- Birth of calves in areas other than designated calving area?; SEC2QN4- How often is non-saleable milk (high risk) fed?). Of variables included in both models, the direction of the associations for significant variable levels are the same, indicating that the direction of the relationship holds true independent of inclusion of the JD herd history. However, in Model B, the estimates for the strength of the association are generally greater than in Model A. Herds were less likely to have a zero count of ELISA positive cows if they were larger herds compared to smaller herds and purchased animals from multiple herds compared to not purchasing animals. Model A additionally specifies that the odds of having no ELISA positive animals were greatly reduced for farms that had previously detected JD clinical or test positive animals in the herd. Herds were predicted to have more ELISA positive cows if they purchased a bull in the last 5 years, if they fed high risk colostrum, and if the hygienic condition of the milking cow environment was poor.

Farms were predicted to have fewer positive cows if calves were fed low risk whole milk compared to milk replacer; but, there was no difference between feeding milk replacer and feeding whole milk from cows without selection for JD or from the bulk tank.

In Model A, additional increased counts of ELISA positive animals were predicted if farms had a history of JD clinical or test positive animals and if more than 10% of calves were born outside of the designated calving area. Model B predicted a reduced ELISA positive count per herd if farms regularly fed non-saleable milk (high somatic cell count, treated, or mastitis milk) to the calves compared to never feeding non-saleable milk (Table 3.3). Interaction terms between feeding non-saleable milk (SEC3QN4) and any of the other milk or colostrum feeding variables

(SEC3QN1 and SEC3QN3) were not significant.

Table 3.2 Mixed logistic regression analyses investigating the association between receiving a recommendation for a specific management area and the RAMP score for that section and Herd ELISA Status (HES) OR1 SE2 p-value ICC3 Recommendation for General Johne’s Disease Risk Management Section 1[categorical (0/1)]4 23.51 4.46 <0.001 0.388 HES5 5.17 1.45 <0.001 Section 1* HES 0.22 0.07 <0.001 Recommendation for Calving Area Risk Management Section 26 1.09 0.009 <0.001 0.245 HES 2.98 1.308 0.013 Section 2* HES 0.97 0.013 0.014 Recommendation for Heifers – Pre-weaned Risk Management Section 37 1.249 0.029 <0.001 0.149 Section 3*Section 3 0.997 0.0004 <0.001 Recommendation for Heifers – Weaned to First Calving Risk Management Section 48 1.09 0.010 <0.001 0.198 Recommendation for Adult Cows Risk Management Section 59 1.31 0.068 <0.001 0.331 Section 5*Section 5 0.996 0.0012 <0.001 HES 0.65 0.14 0.044 1 OR= Odds Ratio 2 SE= Standard Error 3 ICC= intra class correlation coefficient; shows the amount of clustering by veterinarian 4 Section 1 = General Johne’s Disease Risk Management (Range: 0-60 points) 5 HES= Herd ELISA status 6 Section 2 = Calving Area Risk Management (Range: 8-74 points) 7 Section 3 = Heifers – Pre-weaned Risk Management (Range: 7-58 points) 8 Section 4 = Heifers – Weaned to First Calving Risk Management (Range: 4-40 points) 9 Section 5 = Cows – Risk Management (Range: 4-45 points)

Table 3.3 Zero-inflated negative binomial models A and B predicting the number of ELISA positive animals in an Ontario dairy herd. Model A Model B Negative binomial portion Count Ratio Robust SE P-Level Count Ratio Robust SE P-Level SEC1QN2. Have you ever had any JD clinical or test positive animals? Don’t know Referent - - Excluded No 1.08 0.26 0.746 Yes 1.75 0.29 0.001 SEC1QN3_1_3. Purchased bulls in the last 5 years? No Referent - - Referent - - Yes 1.34 0.13 0.003 1.42 0.15 0.001 SEC2QN6. Birth of calves in areas other than designated calving area? Never Referent - - 1-5% of calvings 0.96 0.11 0.691 6-10% of calvings 0.94 0.15 0.687 >10% of calvings 1.41 0.20 0.013 SEC3QN1. Are calves fed low risk cow or artificial colostrum? Colostrum from single low risk cow (own Referent - - Referent - - mother or donor) or artificial colostrum Colostrum from own mother 1.18 0.14 0.163 1.39 0.20 0.023 Colostrum from donor (pooled or frozen), no 1.38 0.21 0.037 1.52 0.26 0.015 selection for JD, fed to 1-10% of calves Colostrum from donor (pooled or frozen), no 0.95 0.12 0.690 1.30 0.21 0.101 selection for JD, fed to >10% of calves SEC3QN3. Are calves fed whole low risk cow milk or milk replacer? Milk replacer for the last two years Referent - - Referent - - Low risk whole milk, selection for JD- or 0.68 0.09 0.004 0.69 0.10 0.010 young cows Whole milk from individual cows, no 0.93 0.14 0.648 1.09 0.19 0.633 selection Bulk tank milk 1.04 0.11 0.742 1.17 0.15 0.212 SEC3QN4. How often is non-saleable milk (high risk) fed? Never Referent - - Once or twice per year (rarely) 0.87 0.13 0.345 Once or twice per month 0.64 0.09 0.002 Weekly 0.68 0.09 0.002 SEC5QN1_2. Milking cow area environmental hygiene? No visible manure Referent - - Referent - - Trace amounts of manure visible, mangers 1.06 0.08 0.484 1.18 0.11 0.079 and water troughs cleaned more than once per month Manure clearly visible OR mangers and water 1.38 0.22 0.047 1.39 0.22 0.038 troughs cleaned less than once per month Extensive manure contamination 1.73 0.42 0.023 2.17 0.55 0.002 Ln(total number tested) Offset Offset Inflated portion Coefficient Robust SE P-Level Coefficient Robust SE P-Level SEC1QN2. Have you ever had any JD clinical or test positive animals? Don’t know Referent - - Excluded No -0.86 0.45 0.057 Yes -22.68 0.37 <0.001 SEC1QN3. Purchased animals in the last 5 years? No Referent - - Referent - - Yes, from 1 herd -0.29 0.32 0.368 -0.13 0.31 0.670 Yes, from multiple herds -0.63 0.27 0.020 -1.20 -.34 <0.001 Ln(total number tested) -0.85 0.23 <0.001 -1.33 0.20 <0.001 Constant 4.77 0.97 <0.001 5.18 0.86 <0.001 Lnalpha -0.399 0.197 0.042 0.27 0.11 0.020

In this study, associations among the RAMP scores, herd ELISA results and RAMP recommendations were investigated. The odds of having a positive HES was >1 for increasing

SEC and the overall RAMP scores, indicating that herds with a higher score were more likely to be HES positive. The number of animals tested was positively associated with the HES and confounded the relationships between some RAMP SEC scores and the HES. This might be due to different management practices in those areas between smaller and larger herds. However, the score for the weaned heifer rearing area (SEC 4) was not associated with the HES. This might be due to a lack of sensitivity of the hygiene scoring for this area or an only negligible relationship with the outcome. The strength of the relationship between SEC 2 and HES is likely 79

overestimated because of the redundancy of questions SEC2QN5 and the JD prevalence (i.e. whether farms were calving JD positive cows in calving area). In SEC 1, question SEC1QN2 was not given a score and the JD herd history was therefore not included in the overall or SEC score through this question.

The relationship between the RAMP SEC scores and SREC was positive but not always linear.

Therefore, herds were more likely to receive a recommendation for a specific management area if they had received a higher score for this area. It was surprising that for the SEC 3 and 4 the probability of receiving a SREC for those areas was not influenced by the HES. Further, for SEC

5 the probability was even lower for HES positive than for HES negative herds. It might be that veterinarians focused their attention on management changes in other areas (SEC 1 and 2) with a more clearly demonstrated link to JD prevention. Therefore, in such cases producers with an

HES positive status may not have been more, or even less, likely to receive a SREC for that area.

Furthermore, a great amount of variability in SREC was explained by the assessing veterinarian.

This might further indicate that there were differences in veterinarians’ preferences in giving recommendations for specific management areas. To the authors’ knowledge, this is the first study investigating the relationship between risk scores for JD transmission and the probability of receiving a recommendation for a management area.

In both ZINB models, purchasing animals from multiple herds was found to be a risk factor for being an ELISA positive herd. This is consistent with previous reports in which purchasing animals was associated with a positive herd status (Wells and Wagner, 2000; Sorge et al., 2012).

However, purchasing animals from one herd was not associated with the HES. It might be that producers that purchased from one herd only, knew the herd and the herd’s disease status and therefore purchased low risk animals compared to producers that purchased from multiple herds.

After accounting for purchasing animals within the last 5 years in the logistic portion of the model, questions about purchasing cows or heifers were no longer significant and only purchasing bulls remained significant in the negative binomial portion of the model. This might indicate that purchasing bulls was associated with additional risks that are not included in just bringing in animals from other herds. It is unclear whether these bulls were used for breeding purposes or were simply beef bulls on the farm. MAP has been isolated from sperm and reproductive organs of breeding bulls (Khol et al., 2010), but transmission of MAP through mating has not yet been reported. Contrarily, Tiwari et al. (2009) described a JD ELISA positive count ratio <1 if the farms purchased bulls within the last year. Those authors suggested that it was a surrogate for an unmeasured herd management variable.

Consistent with previous reports (Muskens et al., 2003; Wells and Wagner, 2000), herd size was a significant risk factor for having at least one ELISA positive animal in the herd. One might argue that due to the imperfect specificity of the test, a small number of animals might test positive, even though they are truly non-infected or negative (false positive). Considering that the herd sensitivity and specificity are dependent on the herd size, herds might be more likely to have at least one test positive cow, the cut-off for HES in this study, if more animals were tested

(Dohoo et al., 2009). Beyond being merely an effect of herd sensitivity and specificity, herd size could have further affected the probability of a herd having a non-zero count. Herd size could be a measure of herd density at calving as outlined earlier (Godkin, 2011). Thus, in a larger herd, cow density in the calving area might be higher and potentially infectious contacts and environmental contamination could be more likely.

Farms were predicted to have fewer ELISA positive animals if they fed low risk colostrum or artificial colostrum compared to feeding colostrum from the dam or a donor cow without selection for JD. This is consistent with Pithua et al. (2009) and Sorge et al., (2012).

A surprising result of these models was that herds had fewer ELISA positive cows if higher-risk practices were recorded for milk-feeding management questions SEC3QN3 and SEC3QN4. This is contrary to observations by Tavornpanich et al. (2008), who identified feeding non-saleable milk as a risk factor for the transmission of MAP. Conversely, according to Nielsen et al. (2008), feeding high SCC milk or bulk tank milk was not associated with a positive ELISA status. The present study is based on a cross-sectional design and causal inferences of the observed associations cannot be drawn. One might even speculate that reverse causation could have occurred in cases where farmers knew their JD status and changed their management accordingly in the desired manner (e.g. feeding milk replacer instead of pooled milk or never feeding non- saleable milk compared to regularly feeding non-saleable milk) in order to reduce the risk of disease transmission on their farm.

Poor milking cow area environmental hygiene was a risk factor for a greater number of ELISA positive cows in the herd. It might be that an unhygienic environment contributes to an infection of adult animals as they enter the milking herd. As the animals get older it is more difficult, but not impossible, for them to get infected with JD. Windsor and Whittington (2010) showed that only 19% of animals infected after 12 months develop a detectable infection and even less develop signs of clinical JD. Several studies summarized by the aforementioned authors reported detection of MAP by culture in intestines or lymph nodes in animals infected at age >24 months.

Similarly, a recent observational study confirmed that infection of adult, MAP unexposed cows was possible after introduction into a JD positive herd (Espejo et al., 2013). Consequently,

environmental contamination through actively shedding animals might contribute to a higher within-herd JD prevalence in the milking herd.

The model predictions for Model A and Model B were very close to the actual observed count probabilities. Therefore, both of the models describe the data appropriately. However, Model B, even though it did not contain the JD herd history, seemed to predict the ELISA prevalence slightly better than Model A which included the herd history. As indicated by the larger model coefficients, Model B placed more emphasis on herd management variables than Model A.

Previous reports have also included the JD herd history in model predictions (Tavornpanich et al., 2008; Johnson-Ifearulundu and Kaneene, 1998; Sorge et al., 2012). However, the current example shows that excluding the herd history variable yielded an equally good model with slightly better predictions. As a practical application of this knowledge, putting less emphasis on the JD herd history when trying to reduce the JD within-herd ELISA prevalence might motivate producers to work on their management practices. It might help to empower producers to change their management rather than “penalizing” them for having had a JD positive herd history in the past. On the other hand, there was a strong positive association between the JD herd history and the HES, as well as the within-herd prevalence in Model A, indicating that this is likely valuable information for potential buyers of animals from this herd.

Not all QNs and SECs were associated with the HES; it would therefore be possible to remove some QNs from the RAMP. However, there might be use in retaining those additional QNs, despite the lack of evidence for their relevance for JD control. The QNs provide a valuable opportunity for the veterinarian to inspect management areas that are not commonly considered during herd health visits and discuss management strategies with the producer (Sorge et al.,

2010). Furthermore, the relative importance of management areas for JD control was accounted

for by increasing or decreasing the number of QNs for the SECs. For example, the calving area had a total of 8 QNs and the pre-weaned calf rearing area 7 QNs, whereas the post-weaning calf rearing area only contained 4 QNs.

Moreover, the strength of the relationship between the answer categories of a QN and the herd

ELISA prevalence might not be reflected in the assigned RAMP QN scores. The relationships are rarely linear and the RAMP QN scores might be misleading in this regard. Thus, when using epidemiological models to describe the relationship between ELISA prevalence and herd management variables with categorical responses, the variables should be offered using the categorical description as opposed to continuous variables using the assigned numerical QN scores.

3.6 CONCLUSION

There was a positive association between the HES and most of the RAMP SEC and the overall

RAMP scores. Purchasing animals from multiple herds was a strong risk factor for being a HES positive farm; however, management practices associated with colostrum management and environmental hygiene were associated with having more animals testing positive on the farm.

The definitive role of purchasing bulls in having more positive animals on the farm is still unknown. Overall, the RAMP is a useful tool for the veterinarian and producer for identifying recommendations for management improvement for JD prevention on the farm.

3.7 ACKNOWLEDGEMENTS

This research was kindly funded by the University of Guelph-Ontario Ministry of Agriculture and Food (OMAF) research partnership. The authors sincerely thank Nicole Perkins and Jamie

Imada for their work and dedication to the OJEMAP. Nicole Perkins coordinated the OJEMAP

and Jamie Imada coded all recommendations into numerical codes. Without their help, it would have been difficult to conduct the program smoothly and successfully.

3.8 REFERENCES

Dohoo, I. R., W. Martin, and H. Stryhn. 2009. Veterinary Epidemiologic Research. 2 ed. VER

Inc., Charlottetown, Canada.

Espejo, L. A., N. Kubat, S. M. Godden, and S. J. Wells. 2013. Effect of Delayed Exposure of

Cattle to Mycobacterium avium subsp paratuberculosis on the Development of Subclinical and

Clinical Johne's Disease. Am J Vet Res 74(10):1304-1310.

Godkin, A. 2011. Johne's Disease: The Ontario Program - Experiences from 2005 to 2011. Pages

73-80 in Western Canadian Dairy Seminar, Advances in Dairy Technology, Edmonton, Alberta,

Canada.

Hendrick, S., T. Duffield, D. Kelton, K. Leslie, K. Lissemore, and M. Archambault. 2005.

Evaluation of Enzyme-Linked Immunosorbent Assays Performed on Milk and Serum Samples for Detection of Paratuberculosis in Lactating Dairy Cows. Journal of the American Veterinary

Medical Association 226(3):424-428.

Johnson-Ifearulundu, Y. J. and J. B. Kaneene. 1998. Management-Related Risk Factors for M. paratuberculosis Infection in Michigan, USA, dairy herds. Prev Vet Med 37(1-4):41-54.

Khol, J. L. J., P. P. Kralik, I. I. Slana, V. V. Beran, C. C. Aurich, W. W. Baumgartner, and I. I.

Pavlik. 2010. Consecutive Excretion of Mycobacterium avium subspecies paratuberculosis in

Semen of a Breeding Bull Compared to the Distribution in Feces, Tissue and Blood by IS900

and F57 Quantitative Real-Time PCR and Culture Examinations. The Journal of Veterinary

Medical Science / the Japanese Society of Veterinary Science 72(10):1283-1283-1288.

Kudahl, A. B., S. S. Nielsen, and S. Østergaard. 2008. Economy, Efficacy, and Feasibility of a

Risk-Based Control Program against Paratuberculosis. Journal of Dairy Science 91(12):4599-

4609.

Long, J. S. and J. Freese. 2006 Regression Models for Categorical Dependent Variables Using

Stata. 2 ed. Stata Press.

Nielsen, S. S. and A. K. Ersbøll. 2006. Age at Occurrence of Mycobacterium avium Subspecies paratuberculosis in Naturally Infected Dairy Cows. Journal of Dairy Science 89(12):4557-4557-

4566.

Nielsen, S. S., H. Bjerre, and N. Toft. 2008. Colostrum and Milk as Risk Factors for Infection with Mycobacterium avium subspecies paratuberculosis in Dairy Cattle. Journal of Dairy

Science 91(12):4610-4615.

Pithua, P., S. M. Godden, S. J. Wells, and M. J. Oakes. 2009. Efficacy of Feeding Plasma-

Derived Commercial Colostrum Replacer for the Prevention of Transmission of Mycobacterium avium subsp paratuberculosis in Holstein Calves. Journal of the American Veterinary Medical

Association 234(9):1167-1176.

Sorge, U. S., K. Lissemore, A. Godkin, J. Jansen, S. Hendrick, S. Wells, and D. F. Kelton. 2011.

Changes in Management Practices and Apparent Prevalence on Canadian Dairy Farms

Participating in a Voluntary Risk Assessment-Based Johne's Disease Control Program. Journal of Dairy Science 94(10):5227-5237.

Sorge, U. S., K. Lissemore, A. Godkin, J. Jansen, S. Hendrick, S. Wells, and D. F. Kelton. 2012.

Risk Factors for Herds to Test Positive for Mycobacterium avium ssp. paratuberculosis-

Antibodies with a Commercial Milk Enzyme-Linked Immunosorbent Assay (ELISA) in Ontario and Western Canada. Can Vet J 53(9):963-970.

Sorge, U., J. Mount, D. F. Kelton, and A. Godkin. 2010. Veterinarians' Perspective on a

Voluntary Johne's Disease Prevention Program in Ontario and Western Canada. The Canadian veterinary journal 51(4):403-403-405.

Tavornpanich, S., W. O. Johnson, R. J. Anderson, and I. A. Gardner. 2008. Herd Characteristics and Management Practices Associated with Seroprevalence of Mycobacterium avium subsp paratuberculosis infection in dairy herds. American Journal of Veterinary Research 69(7):904-

911.

Tiwari, A., J. A. VanLeeuwen, I. R. Dohoo, G. P. Keefe, and A. Weersink. 2008. Estimate of the

Direct Production Losses in Canadian Dairy Herds with Subclinical Mycobacterium avium subspecies paratuberculosis Infection. Can Vet J. 49(6):569-576.

Tiwari, A., J. A. VanLeeuwen, I. R. Dohoo, G. P. Keefe, J. P. Haddad, H. M. Scott, and T.

Whiting. 2009. Risk Factors Associated with Mycobacterium avium subspecies paratuberculosis

Seropositivity in Canadian Dairy Cows and Herds. Preventive Veterinary Medicine 88(1):32-41.

Waddell, L. A., MSc, A. Rajic, J. Sargeant, J. Harris, R. Amezcua, L. Downey, S. Read, and S.

A. McEwen. 2008. The Zoonotic Potential of Mycobacterium avium spp. paratuberculosis: A

Systematic Review. Canadian Journal of Public Health 99(2):145-145-155.

Windsor, P. A. and R. J. Whittington. 2010. Evidence for Age Susceptibility of Cattle to Johne's

Disease. The Veterinary Journal 184(1):37-44.

CHAPTER 4

COMPARING ELISA PREVALENCE, RISK FACTORS AND

MANAGEMENT RECOMMENDATIONS FOR JOHNE’S DISEASE

PREVENTION BETWEEN ORGANIC AND CONVENTIONAL DAIRY

FARMS IN ONTARIO

Laura Pieper1, Ann Godkin2, Ulrike Sorge3, Trevor DeVries4, Kerry Lissemore1, David Kelton1

1Department of Population Medicine, University of Guelph, Ontario, Canada, N1G 2W1

2Veterinary Science and Policy Group, Ontario Ministry of Agriculture and Food (OMAF),

Ontario, Canada, NOB 1S0

3Department of Veterinary Population Medicine, University of Minnesota, St. Paul, USA, MN

55108

4Department of Animal and Poultry Science, University of Guelph, Ontario, Canada, NIG 2W1

Correspondence: Dr. David Kelton; [email protected]

4.1 ABSTRACT

Johne’s disease (JD) is a chronic, infectious disease in cattle. Between 2010 and 2013, a voluntary JD control program was successfully launched in Ontario, Canada, including a Risk

Assessment and Management Plan (RAMP) and JD ELISA testing of the entire milking herd.

Over the last decade, the organic dairy industry has been growing. However, organic farming regulations and philosophies may influence the risk for JD transmission on Ontario organic dairy farms. The aim of this study was to investigate differences in JD ELISA test positive prevalence, risk factors for JD and recommendations for JD prevention between organic and conventional dairy herds in Ontario. RAMP results (i.e. RAMP scores and recommendations) and ELISA

results were available for 2,103 dairy herds, including 42 organic herds. When available, additional data on milk production, milk quality, and herd characteristics were gathered. Organic herds had a similar herd-level, but a higher within-herd JD ELISA test positive prevalence compared to conventional herds if they had at least one positive animal on the farm. Organic farms had lower risk scores in the biosecurity area, and higher scores in the calving and the calf- rearing management areas. After accounting for RAMP score, organic farms received fewer recommendations for the calving management area and more recommendations in the adult cow management area. Three different zero-inflated negative binomial models (A, B, and C) were built. Purchase of animals and the herd size were included in the logistic portion of all three models. Colostrum and milk feeding practices were included in the negative binomial portion of all three models; however, purchase of breeding bulls was a risk factor only in models A and B.

By definition, average bulk tank SCC were only included in models B and C and the presence of non-Holstein breeds was only included in model C. Farm type (organic or conventional) was forced into all three models but was only significant in models A and C. Further research is necessary to investigate the apparent disconnect between risk factors and recommendations on organic dairy farms.

4.2 INTRODUCTION

The organic industry is growing worldwide (Willer et al., 2008). In Canada, the production of organic milk has increased from 100,000 hectoliters (hl) in 2000-2001 to 870,000 hl in 2009-

2010. Likewise, the number of farmers producing organic milk has increased from 65 to 206 between 2000 and 2010 (Agriculture and Agri-Food Canada, 2009). Canadian Organic Standards

(COS) have been implemented in 2006 to regulate the organic dairy industry. Farmers can become certified as organic if they comply with the described management practices in the COS.

Desirable management practices included in the standards focuses on field crop, manure, pasture, and livestock management (Government of Canada, 2006). While some of the organic management practices seem to be favorable for maintaining metabolic health and longevity of dairy cattle (Hardeng and Edge, 2001), the risks arising for infectious disease transmission are rarely investigated.

Johne’s disease (JD) is a chronic, infectious disease of ruminants caused by Mycobacterium avium spp. paratuberculosis (MAP). Subclinically infected cattle may show a drop in milk yield, while clinically affected animals develop profound diarrhea and emaciation. Estimates of the herd level prevalence of JD in Canadian provinces range from 33 to 70% of herds (CHAPTER

1). The annual economic losses have been estimated at $1,196 CAD per 100 cows for the

Canadian dairy industry (Tiwari et al., 2008). In 2010, the Ontario dairy industry launched the voluntary Ontario Johne’s Education and Management Assistance Program (OJEMAP) to limit the effects of JD. Detailed information is provided on the program website (www.johnes.ca) or in Pieper et al. (CHAPTER 2). Briefly, producers were encouraged to conduct an on-farm JD

Risk Assessment and Management Plan (RAMP) with their veterinarian. Until August 2013, dairy farmers were also eligible for a one-time milk or serum JD ELISA testing of the milking herd. The costs for this test were covered by the program provided two additional requirements were met: 1) within 90 days of the test, the producers had to complete a JD RAMP with their herd veterinarian and 2) within 90 days after the test (or before the next calving, whichever occurred first) producers had to dispose of animals with a high positive JD ELISA test result

(HP), while ensuring that those animals did not enter the human food chain or another dairy herd. The RAMP was a standardized one-page questionnaire administered by a trained and registered veterinarian. A handbook was provided to ensure consistency among assessors. The

RAMP was comprised of 5 sections: 1) general JD management (60 points), 2) calving area risk management (80 points), 3) pre-weaned calf area risk management (70 points), 4) post-weaned calf area risk management (40 points), and 5) adult cows risk management (50 points). The maximum score was 300 with higher scores indicating higher risk for JD transmission. The

RAMP was found to be useful for developing a farm specific JD prevention strategy and in predicting the herd-level and within-herd JD ELISA test positive prevalence Pieper et al.

(CHAPTER 3).

For JD, routes of MAP transmission from adult cows to young calves include in utero transmission from dam to calf, and the ingestion of infectious feces, colostrum, and milk

(McKenna et al., 2006). It is generally recommended to remove the calf from the adult cow environment quickly after birth and to provide it with cleanly harvested colostrum and milk from test-negative cows, or with artificial colostrum and milk replacer (Wraight et al., 2000). This latter practice specifically is not consistent with organic farming practices, which encourages bonding between the calf and the cow through prolonged contact periods and suckling of colostrum and milk. While suckling is not a requirement within the COS, natural colostrum and milk has to be provided to the calves (Government of Canada, 2006). In emergency situations, milk replacer from organic sources could be used, but none is currently available in Canada from commercial sources.

Infection with MAP may remain undetected for years. Adult animals between 2.5 and 5.5 years of age have the highest probability of being identified as infected by in vivo diagnostic tests, either indirectly using antibodies against JD or directly based on fecal culture and/or PCR

(Nielsen and Ersbøll, 2006). Furthermore, the probability of detecting JD infection increases with cow age (Jubb et al., 2004). Therefore, since the longevity of cows on organic farms is

greater (Stiglbauer et al., 2013; Hardeng and Edge, 2001), there may be an increased opportunity for infected cows to contaminate the environment and transmit JD to the offspring.

The COS state that breeds and lines of cattle resistant to diseases should be used on the operation

(Government of Canada, 2006). Crossbreeds have been shown to have better reproductive performance compared to purebred Holstein cows (Schaeffer et al., 2011). Thus, organic farms may be more likely to keep minor breeds or crossbreeds. Sorge et al. (2011) showed that Channel

Island breeds had greater odds of being JD positive compared to Holsteins. Consequently, organic farms might have a greater risk of having JD positive animals on their farm if they keep non-Holstein cows.

To date, four studies have commented on JD prevalence on organic farms. Studies by

Ramanantoanina et al. (2012), Zwald et al. (2004), and Sorge (2014) from Canada and the USA, respectively, suggested that the JD prevalence may be lower on organic than conventional farms.

A study from The Netherlands found a similar herd-level and within-herd JD ELISA prevalence on organic and conventional farms. Using a risk-based management assessment, this latter study also reported elevated risk scores in the biosecurity, calving and calf raising areas compared to conventional farms (Kijlstra, 2005). When considering the higher risk scores found in this study, as well as the previously mentioned expectations of organic farming practices, it is surprising that these studies did not find a higher JD prevalence in the investigated populations. However, those studies did not consider herd size when comparing organic and conventional farms.

Commonly, conventional farms are larger than organic farms and larger farms are reported to have higher odds of having JD positive cows (Muskens et al., 2003; Wells and Wagner, 2000).

The objectives of this study were to compare 1) management areas associated with high risk for

JD transmission, 2) farm specific recommendations that were given for JD prevention, and 3) JD

ELISA positive herd-level and within-herd prevalence between organic and conventional dairy herds participating in the OJEMAP.

4.3 MATERIALS AND METHODS

4.3.1 Participation of organic herds With only about 77 organic dairy farms in Ontario (as of September 2011), the number of organic producers participating in the OJEMAP was expected to be low. In order to increase the sample size for organic farms, an advertisement containing a brief description of this project and offering incentives for participation was placed in the respective organic creamery’s newsletters in spring 2012. Organic farmers who had not yet tested through the program were eligible for an additional incentive of $100 CAD to reimburse the cost for the veterinarian administering the

RAMP.

4.3.2 Data From all dairy farms participating in the OJEMAP, information regarding organic status, barn type, milk quality, breed, average age of the herd, and reproductive parameters, all averaged over

12 months during 2012, were accessed via CanWest Dairy Herd Improvement (DHI) and Dairy

Farmers of Ontario (DFO), where available. The information was combined with the dataset used in CHAPTERS 2 and 3, containing 2103 observations accessed from the OJEMAP database. It included information regarding veterinarian conducting the RAMP, the number of tested and the number of JD ELISA test positive animals, RAMP scores, and management recommendations for JD prevention provided to the farmer.

4.3.3 Statistical analysis Descriptive statistics for production, reproduction, milk quality, RAMP scores, and JD test results were generated for conventional and organic herds. Median, inter quartile range (IQR)

and range were reported for most variables. Alternatively, if the median and IQR would have been non-informative (e.g. due to lack of variation), an arithmetic or geometric mean and 95% confidence interval (CI) was displayed. Comparisons of continuous variables were conducted with the Wilcoxon rank-sum test. Where appropriate, Chi-squared or Fisher’s exact test were used for comparison of categorical variables (e.g. keeping a specific breed, comparison of

RAMP question categories).

In a previous publication on the RAMP dataset Pieper et al. (CHAPTER 2), it was shown that

RAMP scores and the likelihood of receiving recommendations were highly dependent on the

RAMP veterinarian. Risk assessment scores and recommendations from the same veterinarian were more similar to each other than to risk assessment scores and recommendations from other veterinarians. Therefore, all logistic, linear, or zero-inflated negative binomial (ZINB) models accounted for assessing veterinarian as a random effect or by using clustered sandwich estimates.

Mixed logistic and linear regression models were built to investigate the associations between organic status and RAMP section scores as well as between organic status and recommendations for JD prevention. Linearity of the relationships was assessed and data transformations, categorizations performed if necessary. For the association between RAMP score and organic status, the logarithmic transformed number of cows tested (lntotnotest) was offered as a confounder to the models. However, the change in coefficient was <30% in all models and lntotnotest was not retained in the final models. For models predicting if a producer received a recommendation for a specific management section, additional to the organic herd status, the

RAMP section score and Herd ELISA status (HES) were offered as confounders and retained if the change in coefficient was >30%. The intraclass correlation coefficient (ICC) for logistic and linear mixed models was calculated as indicated by Dohoo et al. (2009).

Zero-inflated negative binomial (ZINB) models were built to describe the number of JD ELISA positive animals per farm. This type of model was chosen to account for the fact that a zero count of positive animals could arise from two different processes: 1) herds could have zero positive animals because they do not have the disease on the farm, 2) herds could have a zero count of positive animals because they have management practices that reduce the within-herd prevalence so that diagnostic tests cannot find positive animals. The logistic portion of the model is interpreted as the odds of a herd having a zero count of JD ELISA positive animals. The negative binomial portion of the models is interpreted as a count ratio (CR) of ELISA positive animals on a farm (Tiwari et al., 2009). The variable lntotnotest was used as an offset for the negative binomial portion. Variables were offered to the model if they had a probability of p<0.1 in the univariable analyses. Section 1 questions (6 questions), presence of non-Holstein breeds on the farm [yes/no], mean age of the herd in months, and organic status were tested in the logistic and the negative binomial portion of the model. All other RAMP questions (25 questions) were only offered in the negative binomial portion of the model but not in the logistic portion. The variables were retained in the model if the Wald test resulted in p <0.05. Residuals were calculated by subtracting the predicted from the observed counts and they were graphically assessed for outliers.

Three different ZINB models were built. Model A used the full dataset (n=2,103) and only

RAMP variables, the lntotnotest, and the herd type (organic or conventional) as predictor variables. Model B was built similarly to Model A but the milk quality parameters Average Bulk

Tank SCC (BtSCC) and Average Bulk Tank Bacterial Count were additionally offered. Due to the unavailability of this data for all herds, this model contained only 2,069 observations. Model

C was built similarly to Model B, but the presence of non-Holstein dairy cows on the farm was

additionally offered which reduced the number of observations to1,952 (only DHI members).

Because some data (e.g. BtSCC, breed) were not available for all herds, a comparison between herds with and without data available was conducted.

The statistical analyses were performed using the computer program STATA 13.0 (StataCorp

LP, Texas, USA). A probability of p<0.05 was considered significant.

4.4 RESULTS

A total of 42 organic herds (54.5% of all Ontario organic herds) participated in the OJEMAP. On average, organic farms had a higher average age of the cows, lower milk production and higher bulk tank and test day somatic cell counts (SCC) compared to conventional dairy herds. The herd size was numerically lower on organic farms but the difference was not significant (Table

4.1). Additional comparisons of milk quality characteristics are displayed in Appendix III.I.

Organic farms also more frequently kept cows of non-Holstein breed on their farm. Significant differences were observed in the proportion of farms having Brown Swiss, Jersey, and Milking

Shorthorn cows. Of all farms keeping purebred Holstein cows, the within-herd prevalence of this breed was lower on organic compared to conventional farms (Table 4.2). Barn type data were available for 1,918 conventional and 41 organic herds. Overall, there were no differences in frequencies of different barn types between organic and conventional farms (p=0.142; Appendix

III.II).

Organic farms tested numerically fewer cows compared to conventional farms. The average numbers of ELISA positive and HP cows, and the within-herd ELISA prevalence of positive and

HP cows, were similar between farming types (Table 4.3). There were no differences in herd- level ELISA positive prevalence in organic and conventional herds when using either of three common definitions of a positive herd; 1 or more ELISA positive animals (26.2% vs. 27.2%,

respectively; p=0.88), 2 or more ELISA positive animals (19.0% vs. 12.3%, respectively; p=0.19), or 1 or more ELISA HP animals (7.1% vs. 6.2%, respectively; p=0.79). When comparing HES positive herds, organic herds had a higher within-herd prevalence of ELISA positive animals compared to conventional herds (Median: 4.2% vs. 2.3%, p=0.03). Due to the bimodal nature of the RAMP Section 1 score, the variable was dichotomized. The proportions of herds that did not purchase animals in the last 5 years and therefore received a section 1 score of zero were 47.6% and 27.1% for organic and conventional herds, respectively.

After accounting for the RAMP veterinarian, organic farms were less likely to receive a section 1 score >0 (OR=0.38; p=0.003), and received on average a 7.2 points higher score in section 2

(p<0.001), and a 3.9 points higher score in section 3 (p=0.008) compared to conventional farms.

The other section scores and the overall RAMP score were not associated with organic status.

The assessing veterinarian had a great influence on the variability of RAMP section scores for sections 2 to 5, and the overall RAMP score (ICC: 0.23 to 0.32) but not on section 1 score (ICC:

0.04; data not shown).

Significant results of simple chi-square comparisons of RAMP scores in organic and conventional herds, disregarding herd size or assessing veterinarian, are displayed in Table 4.4. It was revealed that organic farms were less likely to purchase animals or cows within the last 5 years. On organic farms, calves were more likely to be born outside the designated calving area, to nurse the cows, to stay with the dam for a prolonged period of time after birth, and to be housed in larger groups pre-weaning. Pre- and post-weaning heifers on organic farms were more likely to be housed close to cows or exposed to cow manure (Table 4.4). Further RAMP score comparisons between organic and conventional farms are displayed in Appendix III.III.

After accounting for section score and HES as fixed effects, and assessing veterinarian as a random effect, organic farms were less likely to receive recommendations for Section 2

(OR=0.41; p=0.037) and tended to be less likely to receive recommendations for Section 3

(OR=0.48; p=0.053). Conversely, they were more likely to receive recommendations for Section

5 (OR=2.7; p=0.048; Table 4.5). A comparison between specific recommendations given to organic and conventional producers is provided in Appendix III.IV.

The final ZINB models are displayed in Table 4.6. Overall, the direction and strength of associations of variables included in all models were very similar. The variables Section 1

Question 3 (S1Q3 Purchased animals in the last 5 years?) and lntotnotest were included in the logistic portions of all 3 models. Owners who purchased animals from multiple herds during the last 5 years were less likely to have zero positive animals on their farm compared to owners that did not purchase animals (closed herds). However, producers purchasing animals from only one farm were not more likely to have at least one positive animal compared to closed herds.

Furthermore, larger herds were less likely to have a zero count of positive animals on the farm.

The variables S3Q1 (Are calves fed low risk cow colostrum or artificial colostrum?), S3Q3 (Are calves fed whole low risk cow milk or milk replacer?), and S3Q4 (How often is non-saleable

(high risk) milk fed?) were included in all models. Herd type (organic or conventional) was retained in the model as the variable of interest. Herds that fed colostrum considered to be of higher risk (e.g. pooled colostrum) were more likely to have a higher number of test positive animals on the farm. For the variables S3Q3 and S3Q4, CRs below 1 were observed, indicating for example that herds that fed low risk whole milk from JD negative or young cows had fewer

JD positive cows compared to herds feeding milk replacer. Similarly, herds which never fed non-

saleable milk (waste milk) had a higher number of JD test positive animals compared to herds that fed non-saleable milk.

According to Model A and Model C, organic farms have about a 2-fold increase in the number of

JD positive animals compared to conventional farms of the same size. In Model B, there was a tendency towards organic herds having an increased CR (p=0.073). Higher BtSCC was a risk factor in Model B and Model C with a CR of 1.002 for each 1,000 cell increase in BtSCC. In other words, a herd with a BtSCC of 300,000 cells/ml over 12 months would have a 1.5 times higher number of JD positive animals than a herd with BtSCC of only 100,000 cells/ml during that year. Additionally, herds keeping non-Holstein dairy cows had 1.6 times more JD test positive animals compared to farms keeping solely Holstein dairy cows.

While S1Q3_1_3 (Purchased bulls in the last 5 years?) was included in Model A and Model B as a predictor variable, it was not retained in Model C, potentially due to the difference in herd characteristics of herds included in Model C compared to herds excluded. Herds (both organic and conventional) from whom no DHI and BtSCC information was available (excluded in Model

C; n=151) more often used serum ELISA (28% vs. 1%, respectively; p<0.001), more often were

HES positive (44% vs. 26%, respectively; p<0.001), had a higher within-herd ELISA test positive prevalence (2.0% vs. 0.8%, respectively; p<0.001), and more often purchased bulls in the last 5 years (32% vs. 15%, respectively; p<0.001) compared to included herds. However, there was no difference in the proportion of organic herds in excluded and included herds (1.3% vs. 2.1%, respectively; p=0.4).

4.5 DISCUSSION

This study investigated potential differences in herd characteristics, JD ELISA test positive prevalence, risk factors and recommendations for JD control between organic and conventional

100

Ontario dairy herds. Organic farms were found to be different from conventional farms in various herd characteristics, some of which are potentially very important for JD prevention. A different pattern of recommendations for JD prevention was observed for organic farms compared to conventional farms.

The number of organic farms in Ontario was low, resulting in a low number of organic herds participating in the OJEMAP. This might have limited the ability to detect differences in risk factors, ELISA results and recommendations between organic and conventional herds. To overcome this potential lack of power, an incentive was offered to organic producers to increase their participation. This might have biased the results of this study by increasing participation of producers that potentially had a different ELISA prevalence compared to herds that participated without an additional incentive (Kelton et al., 2014). However, the proportion of Ontario organic herds (54.5%) and all Ontario dairy herds (51.8%) participating in the OJEMAP was similar, indicating that a potential bias might have been negligible.

Organic producers were less likely to have purchased livestock in the last 5 years compared to conventional producers. This might be due to the organic philosophy of minimizing the importation of external resources onto the farm, or due to lack of the availability of organically raised livestock. These results are similar to the observations by Stiglbauer et al. (2013) and

Stonehouse et al. (2001). Due to their lower risk scores for biosecurity measures (Section 1), organic farmers received fewer recommendations in this area.

Similarly to a study from The Netherlands (Kijlstra, 2005), organic farms in our study had higher risk scores in the calving and pre-weaned calf rearing section, indicating a greater risk for JD transmission. This was not surprising since organic producers often promote closer and longer cow-calf contact and subsequent group housing of pre-weaned calves. However, these practices

101

represent ideals of organic farming rather than practices required by the COS. The COS specifically state that pre-weaned calves should be able to see, hear, and smell each other and that they should be group housed after weaning (Government of Canada, 2006). In spite of the higher risk for JD transmission, organic farms were less likely to receive management recommendations for JD prevention in the calving and pre-weaning calf sections compared to conventional farms, after accounting for the section score. It might have been that veterinarians refrained from recommending management change in the calving and calf rearing areas due to their lack of awareness of the specific requirements for organic certification. Conversely, organic producers might have stated that they were not going to comply with recommendations for those areas (e.g. separating the calf from the dam) and therefore other recommendations that the veterinarian and the farmer agreed on were documented for implementation.

The variables average age and the proportion of cows in the third and higher lactation were not significant predictors in either of the models for herd-level or within-herd JD prevalence. This was surprising since, based on the biology of the disease (e.g. long incubation period and greater likelihood of older cows being detected by diagnostic tests (Nielsen and Ersbøll, 2006)) and the findings by Tiwari et al. (2009), the authors of the current study expected a greater within-herd or herd-level prevalence as average age of the herd increased. It might have been that the measure averaged for the herd and over 12 months was too imprecise to detect differences in prevalence or that reduced longevity of affected animals masked the effects of age on detection of positive animals (survival bias). It might also be that known associations at the cow-level do not translate to associations at the herd-level (ecological fallacy).

To the authors’ knowledge, this is the first study to document the JD prevalence in organic herds concurrently with conventional herds based on ELISA test results. The herd-level JD ELISA

102

positive prevalence in this study was similar in conventional and organic dairy herds. This is in agreement with the study by Kijlstra (2005) but not with Ramanantoanina et al. (2012) and

Zwald et al. (2004). The latter studies found a reduced herd level prevalence in organic herds compared to conventional herds. In Zwald et al. (2004), organic herds were significantly smaller than conventional herds and the farmer reported JD prevalence was lower. In Kijlstra (2005) and

Ramanantoanina (2012), the test results were compared with historical data from conventional herds and it is unclear how comparable to each other those estimates of prevalence were.

Nevertheless, considering that organic farms had a reduced risk of introducing JD onto the farm by purchasing cattle (lower Section 1 score), it is noteworthy that the JD herd-level prevalence was not different on organic and conventional farms. One might speculate that organic farms were affected by JD before converting to organic farming and the infection remained in the herd after the conversion. However, it might also be that due to high risk practices, low prevalence infected farms perpetuated the disease even though the risk of introduction of new MAP infection was lower.

The variable purchasing a bull in the last 5 years was a significant predictor variable in Models A and B but not in Model C. It might be that after exclusion of many herds in Model C, there was a lack of power to detect the importance of this variable. However, it might also be that purchasing a bull was a surrogate measure for another unmeasured herd characteristic for those herds excluded herds from this analysis (Tiwari et al., 2009). On the other hand, case reports of MAP infected bulls described shedding of MAP with feces and semen (Khol et al., 2010), suggesting a potential risk for MAP transmission. Further research is needed to determine whether there is a direct influence of a herd bull on MAP transmission.

103

The number of ELISA positive animals in the herd increased if non-Holstein dairy cows were kept on the farm. A higher odds of non-Holstein cows being JD ELISA positive was also reported by Sorge et al. (2011). Including a question about the breed in the RAMP might help to more precisely estimate the risk for JD test-positivity. On the other hand, the prevalence of non-

Holstein cows within the herds was commonly very low and the higher JD prevalence might not actually reflect a higher JD prevalence among non-Holstein cows. The presence of non-Holstein cows might similarly be a surrogate measure for other unmeasured herd management variables that those herds have in common.

Larger herds were found to be more likely to have at least one JD ELISA test positive animal on the farm. This is consistent with findings by Muskens et al. (2003) and Wells and Wagner

(2000). While herd size is a risk factor that potentially remains unchanged, it might be that including a measure of herd size in the RAMP could help in estimating the risk for JD transmission between herds and in implementing additional prevention efforts on larger farms.

In Models B and C, BtSCC was included as a predictor variable. This variable could serve as a surrogate measure for a number of herd management characteristics including environmental hygiene and milking hygiene. It might be a more objective and long term measure than environmental hygiene scoring performed during the risk assessment. Hygiene scoring is prone to bias due to day-to-day variation in cleanliness and inconsistencies in scoring by different assessors. Therefore, it might be that accuracy of the assessment could be improved by including a long term measure such as BtSCC in the RAMP.

For question S3Q3 (Are calves fed low risk whole milk or milk replacer?) veterinarians marked the lowest risk category (Calves are fed milk replacer since 2 years) on a total of 14 organic farms. Since organic milk replacer is not commercially available, it is likely that veterinarians

104

did not mark this question according to the guidelines in the handbook provided. The marking regimen might be confusing for veterinarians due to the similarity of this question to the question

S3Q1 (Are calves fed low risk cow or artificial colostrum?) but a different categorization of the risk (e.g. lowest risk category: Colostrum from single low risk cow (own mother or donor) or artificial colostrum). It might be that veterinarians would benefit from a regular refresher of the assessment procedures or that those categorizations should be revised to make assessment easier and more consistent.

4.6 CONCLUSION

The herd-level JD ELISA positive prevalence was similar in organic and conventional herds, but the within-herd prevalence was higher in affected organic herds. The presence of non-Holstein cows on the farm was a risk factor for an increased number of ELISA positive animals.

Therefore, breed distributions should be included in the RAMP to better assess the risks for JD transmission on organic and conventional farms. There was a different pattern of high risk management practices on organic compared to conventional farms. However, the recommendations for JD prevention that organic producers received after the RAMP might not have reflected those different risk patterns; especially, in areas where organic farming regulations or ideals are a major influence on management practices. Further research is necessary to investigate the relationship between organic farmers and their veterinarians to find causes for the different patterns in recommendations and to suggest risk reducing management strategies for organic farmers in areas influenced by organic farming regulations.

4.7 ACKNOWLEDGEMENTS

This study was funded through the University of Guelph-OMAF Research Partnership.

105

4.8 REFERENCES

Agriculture and Agri-Food Canada (AAFC). 2009. Certified Organic Production Statistics for

2009/?id=1312385802597#a3

Dohoo, I. R., W. Martin, and H. Stryhn. 2009. Veterinary Epidemiologic Research. 2 ed. VER

Inc., Charlottetown, Canada.

Government of Canada. 2006. CAN/CGSB-32.310-2006 Organic Production Systems - General

Principles and Management Standards.

Hardeng, F. and V. L. Edge. 2001. Mastitis, Ketosis, and Milk Fever in 31 Organic and 93

Conventional Norwegian Dairy Herds. Journal of Dairy Science 84(12):2673-2679.

Jubb, T. F., E. S. G. Sergeant, A. P. L. Callinan, and J. W. Galvin. 2004. Estimate of the

Sensitivity of an ELISA used to Detect Johne's Disease in Victorian Dairy Cattle Herds.

Australian Veterinary Journal 82(9):569-573.

Kelton, D., N. Perkins, A. Godkin, G. MacNaughton, R. Cantin, K. Hand, and M. Watters. 2014.

Comparison of Participants and Non-Participants in a Voluntary Johne's Disease Control

Program in Ontario, Canada. Page 108 in 12th International Colloquium on Paratuberculosis,

Parma, Italy.

Khol, J. L. J., P. P. Kralik, I. I. Slana, V. V. Beran, C. C. Aurich, W. W. Baumgartner, and I. I.

Pavlik. 2010. Consecutive Excretion of Mycobacterium avium subspecies paratuberculosis in

106

Semen of a Breeding Bull Compared to the Distribution in Feces, Tissue and Blood by IS900 and F57 Quantitative Real-Time PCR and Culture Examinations. The Journal of veterinary medical science / the Japanese Society of Veterinary Science 72(10):1283-1283-1288.

Kijlstra, A. 2005. No Difference in Paratuberculosis Seroprevalence Between Organic and

Conventional Dairy Herds in The Netherlands. In: Hovi, M.; Zastawny, J. und Padel, S. (Hrsg.)

Proceedings of the 3rd SAFO Workshop; Enhancing Animal Health Security and Food Safety in

Organic Livestock Production., University of Reading, Reading, UK, S. 51-56.

McKenna, S. L., G. P. Keefe, A. Tiwari, J. VanLeeuwen, and H. W. Barkema. 2006. Johne's

Disease in Canada Part II: Disease Impacts, Risk Factors, and Control Programs for Dairy

Producers. Can Vet J 47(11):1089-1099.

Muskens, J., A. R. W. Elbers, H. J. Van Weering, and J. P. T. M. Noordhuizen. 2003. Herd

Management Practices Associated with Paratuberculosis Seroprevalence in Dutch Dairy Herds.

Journal of Veterinary Medicine, Series B 50(8):372-377.

Nielsen, S. S. and A. K. Ersbøll. 2006. Age at Occurrence of Mycobacterium avium subspecies paratuberculosis in Naturally Infected Dairy Cows. Journal of Dairy Science 89(12):4557-4557-

4566.

Ramanantoanina, F., D. Francoz, G. Côté, O. Labrecque, A. C. Gagnon, G. Beauchamp, G.

Fecteau. 2012. Seroprevalence of Mycobacterium avium subsp. Paratuberculosis, Neospora caninum, Bovine Viral Diarrhea and Infectious Bovine Rhinotracheitis Viruses in Québec

Organic Herds. in XXVII. World Buiatrics Congress, Lissabon, Portugal.

107

Schaeffer, L. R., E. B. Burnside, P. Glover, and J. Fatehi. 2011. Crossbreeding Results in

Canadian Dairy Cattle for Production, Reproduction and Conformation. The Open Agriculture

Journal 5:63-72.

Sorge, U. S. 2014. An Overview Over the Organic Dairy Industry in Minnesota. in 117th Annual

Meeting of the Minnesota Veterinary Medical Association, Minneapolis, MN, USA. February 6-

8, 2014.

Sorge, U. S., K. Lissemore, A. Godkin, S. Hendrick, S. Wells, and D. Kelton. 2011. Associations

Between Paratuberculosis Milk ELISA Result, Milk Production, and Breed in Canadian Dairy

Cows. Journal of Dairy Science 94(2):754-761.