| dc.description.abstract | Infectious diseases of livestock can cause substantial production losses and have detrimental impacts upon human health, and animal health and welfare. To limit the impact of diseases, understanding more about the dynamics of transmission can assist in the control and prevention of infectious disease. In particular, understanding infection transmission on networks, ‘network epidemiology’, offers a flexible approach, incorporating between-host heterogeneity in potentially infectious contacts drawn from empirical study of interactions among individual animals, or among farms. Trading animals and optimising productivity are vital to the commercial viability of farms, however they necessarily involve compromises in biosecurity, animal health, and welfare. Better understanding of the relationships among these multiple factors might facilitate the development of sustainable livestock industries that are more resilient to disease outbreaks. In this thesis I examine cattle interactions at two spatial scales, first at a national-level by studying the trading connections among farms, and then at a finer scale by analysing the social interactions among cattle. First, I introduce the concept of superspreaders, hosts that generate many more secondary infections than the rest of the population, and evaluate evidence for the notion that some farms might act as superspreaders of infection. I utilise the example of bovine tuberculosis (bTB) to illustrate this concept and find that farms might act as superspreaders in three main ways; first, via exceptional trading between farms, second, by factors that facilitate high within-herd transmission and trading of high-risk animals, and third, by harbouring undetected infection for long periods. I find mechanisms that align with all three processes in the cattle industry in Great Britain that might allow superspreader farms to contribute to the current bTB epidemic. At a national level, I describe cattle movements among farms over time, finding that some farms consistently act as ‘hubs’ in trading networks, functioning in a similar way to markets, in that they are highly connected to other farms by many direct trades. Utilising the temporal network measure of ‘contact chains’, I quantify the farms that represent potential sources of infection (ingoing contact chains) and the potential farms that a farm might infect over 1 year periods. Farms divide into two groups: those with very few connections (less than 10 farms) that are relatively isolated from the network, and those with very many connections (more than 1000 farms) that are highly connected within the network. I find that a substantial number of farms have over 10,000 farms in both their ingoing and outgoing contact chains, such that, if infected, they might potentially act as superspreaders by being more at risk of both acquiring and spreading infection. Building on my previous analysis, I then characterise the ‘source farms’ in the ingoing contact chains, in terms of their location and bTB history. I find that after controlling for previously-established risk factors for bTB, having more source farms in areas of higher bTB risk in the ingoing contact chain increases the odds of a bTB incident on the root farm, whilst having more source farms in lower risk areas is associated with lower odds of a bTB incident on the root farm. At a finer scale of contacts among animals, I explore interactions among dairy cattle in multiple herds using automated proximity sensors and GPS devices. When aggregated over long periods, cattle interactions appear dense and unstructured, however finer time spatial and temporal perspectives revealed structure and variation in contacts. Herds in our study had variable grazing and housing access, allowing us to determine that cattle interact with more other cows, for longer time periods when they are in buildings compared to contacts at pasture. Cattle exhibited heterogeneity in their number and duration of contacts, and although the majority of cattle interacted more equally with other cows, a small proportion of cows in each group showed evidence of stronger social ties. Next, I consider associations between social interactions, production, and health. I review the existing literature on social parameters such as dominance rank and re-grouping of cattle, and find inconclusive outcomes regarding their impact on milk yield and somatic cell count, an indicator of udder health. I perform my own analysis to examine the relationship between the time cows spend with other cows, milk yield and somatic cell count, and do not find a statistically significant relationship. In considering social preference, cows that had experienced the same number of lactations were more likely to interact, but cows spending more time with cows in the same lactation did not appreciably affect their milk yield or somatic cell count. Finally, I draw together the findings of this thesis and reflect on how the identification of higher-risk farms might be useful in the control of livestock infections, and specifically bTB in Great Britain. I conclude that network analysis is a valuable tool to study the interactions of cattle and cattle farms, identifying unique opportunities for targeted approaches to disease control. | en_GB |
