The sheep industry is currently facing increasing social demands for assurances of good animal welfare, which indicates that assessing and addressing the welfare of sheep is critical if the current ‘social license to farm sheep’ is to be maintained. Farmers are key players in improving sheep welfare because they are responsible for the care of their animals and make the management decisions on their farm. Hence, a better understanding of the main factors underpinning farmer behaviour is important. According to the Theory of Planned Behaviour, farmer attitudes towards sheep management are likely to influence their behaviour in terms of the implementation of improved management practices. Subsequently, farmer management behaviour may impact on the welfare of their sheep. The aim of this thesis was therefore to examine the relationships between farmer attitudes, management behaviour and sheep welfare outcomes. To examine these relationships, animal welfare assessments and questionnaire interviews with farmers were conducted. A total of 6200 ewes (aged 2-5 years) and 32 farmers were sampled across Victoria, Australia. Farms were visited at mid-pregnancy and weaning, and the ewes were examined using six animal-based measures: body condition score (BCS), fleece condition, skin lesions, tail length, dag score and lameness. In addition, the number of ewes that needed further inspection/care (such as sick or injured sheep) was recorded and reported to the farmers. The welfare of the ewe flocks, based on the six indicators measured, was good overall. However, individual welfare compromise was considered significant. There were 185 (3.0%) ewes needing further care and cases were identified in all farms. Main reasons for further inspection/care were lameness or foot-related issues, BCS ≤ 2 and active dermatophilosis or broken wool. Farmer attitudes to sheep and management were in general positive and welfare was considered an important aspect of farm productivity. In terms of management, it was found that farmers relied mostly on visual monitoring to assess the flocks and pasture quality and quantity. However, close inspection of the flock and more detailed practices such as body condition scoring of sheep, veterinary interventions or keeping accurate records were less common activities. Indeed, mortality rates were frequently underestimated. Positive farmer attitudes were associated with positive management behaviour (active management style), and positive management behaviour was associated with positive ewe welfare outcomes. Farmers with an active management style had fewer ewes in need of further inspection/care after both welfare assessments (mid-pregnancy and weaning). Main results indicated that farmers were more likely to perform an activity (management behaviour) if they perceived the activity was valuable (behavioural attitudes) and if they perceived the activity could be realised (perceived behavioural control). The results of this thesis demonstrate that there is an opportunity to create change in farmer management behaviour and potentially improve sheep welfare via education programs targeting attitudinal change. The results reported in this thesis provide what is believed to be the first comprehensive study investigating the on-farm welfare of extensively managed ewes and the farmer-sheep relationship in extensive systems. Controlled trials to assess if education programs or other interventions are able to improve farmer attitudes and management behaviour, as well as sheep welfare, would allow causality to be investigated and would be a valuable next research step following on from the current work.

Traditionally, urinary tract infections (UTIs) have been categorised as either uncomplicated or complicated in veterinary medicine, with treatment differing for the two categories. In human medicine, there is an additional category: asymptomatic bacteriuria, which is the presence of bacteriuria without symptoms of infection. Escherichia coli is the most common bacterial species involved in UTIs in dogs. Clinical signs can be absent in dogs with complicated UTIs, and this has been likened to asymptomatic bacteriuria in people and has been termed subclinical bacteriuria (SBU). Treatment recommendations for SBU in dogs have been adapted from human recommendations. Many E. coli strains are resistant to multiple antibiotics and uropathogenic E. coli possess virulence factor genes that facilitate overcoming host defence mechanisms. These E. coli commonly belong to phylogenetic groups B2 and D. Some studies suggest that human E. coli isolated from asymptomatic bacteriuria differ from those causing clinical UTI. While the virulence factor genes and phylogeny of canine urinary E. coli isolated from UTIs are well characterised, little is known about virulence factor genes and phylogeny in E. coli isolated from SBU. Furthermore, these genomic characteristics have not been studied in detail in multi-drug resistant (MDR) canine urinary E. coli. Thus, the pathogenic potential of canine urinary E. coli is not well described and the benefit of antibiotic therapy in SBU and MDR infections is unknown. This study used whole genome sequencing to characterise 47 E. coli isolated from dogs with SBU and 67 E.coli from dogs with clinical UTI in terms of their phylogeny and virulence factor gene profile. From those strains, the 15 MDR strains were characterised. When the phylogeny and virulence factor gene profile of E. coli isolated from dogs with SBU were compared with clinical UTIs, results showed that most clinical UTI and SBU E. coli belonged to phylogenetic group B2. The virulence factor gene profile was similar between the two groups, and no association was found between them and the 83 virulence factor genes analysed. Many of the MDR E. coli belonged to phylogenetic group B1 and these isolates possessed fewer virulence factor genes than non-MDR E. coli. Based on the results of this study, it was concluded that phylogeny and the presence of virulence factor genes do not influence the manifestation of clinical disease. Host immunity and rather than presence and the expression of or mutations of virulence factor genes may have a role in the development of clinical disease. ii MDR E. coli have fewer virulence factor genes than non-MDR isolates, with MDR isolates commonly belonging to commensal phylogenetic groups. This suggests that treatment of MDR isolates is not always indicated because they tend to have commensal bacterial phylogeny, which can have implications for reducing the development of antimicrobial resistance. While whole genome sequencing is an accurate method for determining phylogeny and the presence of virulence factors, it is unable to easily differentiate the pathogenicity potential of urinary E. coli and therefore treatment recommendations cannot be made based on this technique.

Chronic enteropathy (CE) is an umbrella term used in dogs to describe a group of diseases with different aetiologies, characterised by chronic gastrointestinal signs. These diseases are clinically classified according to treatment response as food-responsive (FRE), antibiotic-responsive (ARE), and immunosuppressant-responsive enteropathies (IRE). The first part of this PhD thesis prospectively describes the features of CE that are commonly seen in dogs presenting to a referral centre in Australia; information that has not been available previously. We found that similar to other countries, most dogs with CE are food-responsive, followed by antibiotic-responsive with a minority of immunosuppressant-responsive. Furthermore, our study raised concerns about prolonged antibiotic treatment for dogs with ARE. Firstly, most of these dogs do not respond to treatment completely for prolonged periods (as opposed to dogs with FRE that do), raising the question about the real benefit of antibiotic treatment. Secondly, half of the dogs with ARE required long-term or pulse treatment with antibiotics, which raises concerns about development of bacterial resistance. Our findings highlight the need to find alternative treatment for dogs with ARE in view of the poor long-term outcome. Although most dogs with FRE had long-term remission, adequate dietary trials were not performed until reaching the referral setting. This indicates that better education of general veterinary practitioners about the importance of performing adequate diet trial is needed to improve early disease remission in these dogs. The next focus of the research was to evaluate the role of macrophages in CE; this was achieved by using two macrophage markers: calprotectin and cluster of differentiation 163 (CD163) in immunohistochemical examination. Both immunohistochemical markers highlighted two different populations of macrophages in our intestinal biopsy specimens. Overall the number of CD163 positive cells was higher than calprotectin positive cells both in crypts and villi. Dogs with FRE and IRE had a decreased CD163:calprotectin ratio compared to healthy dogs with an increase in the ratio after treatment. Our results suggest that there is an imbalance in macrophage populations in dogs with FRE and IRE, with partial resolution following clinical response characterised by an increase in the ratio CD163:calprotectin. Interestingly, dogs with ARE not only have a poor long-term response, but also have different macrophage populations from dogs with FRE and IRE; and in fact, are very similar to healthy dogs without change in their macrophage populations with treatment response. These results suggest that macrophages play a role in the pathogenesis of FRE and IRE dogs with normalisation of macrophage populations with treatment response. The CD163 receptor is cleaved during macrophage activation and is released into the circulation as a soluble form. In view of the decreased number of CD163 cells in the intestine of dogs with FRE and IRE at diagnosis (and subsequent increase with clinical remission), we wanted to determine if soluble CD163 could be detected in dog serum, and therefore potentially serve as a biomarker. Two different ELISAs were tested and although one of them showed some signal, further testing of the antibodies used in the assay did not support that the signal was specific for CD163. With this experiment, we were not able to quantify soluble CD163 in dogs, but this molecule retains potential as a biomarker for diagnostic and monitoring purposes in CE as well as in other diseases characterised by macrophage activation. Biomarkers of systemic inflammation were also assessed in the same cohort of dogs and we showed that serum IL-6 decreased in dogs with CE after resolution of clinical signs. Similarly to soluble CD163, cytokines might play a role in further differentiating between the different causes of CE for prognostic and therapeutic purposes. Future studies are needed to assess these cytokines in a larger cohort of dogs to be able to study differences between FRE, ARE, and IRE, and further assess their role as biomarkers. Finally, we studied cytokine production by lymphocytes or monocytes in the peripheral blood of healthy dogs, and the effect of different immunosuppressive treatments on cytokine production. Differential activation of lymphocytes or monocytes can easily be achieved by using specific activators in whole blood. The advantage of this technique is that there is minimal handling of the cells with less risk of iatrogenic activation. Cytokine production was affected by cyclosporine and prednisolone, but not by mycophenolate, leflunomide, or azathioprine. Cyclosporine inhibited production of tumour necrosis factor (TNF), interferon gamma and IL-10 by lymphocytes whereas prednisolone inhibited TNF production by both lymphocytes and monocytes. Our findings suggest that this methodology can be used to monitor dogs treated with both drugs concurrently – although this needs to be further assessed with future studies. Future studies highlighted by our research suggest more in-depth assessment of serum cytokines as biomarkers for dogs with CE not only for monitoring purposes, but to determine if different patterns of cytokines can be useful to refine the classification of CE. Similarly, whole blood stimulation can be used to better assess underlying priming of the immune system and to monitor treatment response. Finally, our findings suggest that macrophages play a significant role in the pathophysiology of CE in dogs, particularly in FRE and IRE, but additional work is required to better understand their function in CE.

Advances in veterinary medicine have led to the development of further techniques to provide analgesia for our patients. Ovariohysterectomy is a commonly performed surgery in female dogs which causes significant pain. Multimodal analgesia often combining opioids, non-steroidal anti-inflammatories and local anaesthetic techniques is implemented to treat this type of pain. The transversus abdominis plane (TAP) block is a local anaesthetic technique that provides analgesia to the afferent nerves supplying the parietal peritoneum, muscles and skin of the anterior abdominal wall. The TAP block is utilized in human anaesthesia and has been shown to reduce post-operative pain scores and opioid consumption. The TAP block has potential to provide local anaesthesia to the mid-caudal abdomen for dogs undergoing ovariohysterectomy. The initial pilot study was a cadaveric study carried out in 5 dogs with an ultrasound guided two-point TAP injection performed in each hemiabdomen. The study identified that a two-point TAP injection delivered consistent dye dispersion to adequately stain branches of thoracic nerve T13 and lumbar nerves L1, L2 and L3. It concluded that this technique should be assessed in vivo to evaluate the analgesic efficacy in mid to caudal abdominal surgeries. A subsequent randomized controlled trial assessed the two-point ultrasound-guided ropivacaine TAP block for dogs undergoing ovariohysterectomy, with the hypothesis that dogs receiving the ropivacaine TAP block would have a lower minimum alveolar concentration (MAC) isoflurane at skin incision and a lower opioid rescue analgesic requirement post-operatively. The results concluded that the estimated population MACiso for the treatment group was lower than the control group but that this was not significant. There was also no significant difference in pain scores between the groups as measured by the Glasgow composite pain scale short form and mechanical nociceptive threshold testing. The proportion of subjects receiving rescue analgesia in the treatment group was lower than the control group, but this was not significant. The study concluded that the two-point TAP block did not provide significant MAC reduction or additional intra-operative or post-operative analgesia compared with the control for dogs undergoing ovariohysterectomy. Multiple confounding factors such as the administration of morphine as premedication and the low concentration of local anaesthetic used, ropivacaine 0.2%, could have contributed to these findings. Although the cadaver study identified that a two-point TAP injection delivered consistent dye dispersion to thoracic nerve T13 and lumbar nerves L1, L2 and L3, the technique when performed with 0.2% ropivacaine in dogs undergoing ovariohysterectomy did not provide significant additional MACiso reduction or intra-operative or post-operative analgesia compared with the control for dogs. Further assessment of the toxic dose of ropivacaine in dogs and the minimum effective concentration of ropivacaine for TAP blocks in dogs needs to be evaluated in order to further develop and assess this technique.

Prior to the studies detailed within this thesis butorphanol was the only opioid registered for use in cats in Australia. There was a need for further clinical studies to be conducted in cats to evaluate the perioperative analgesic efficacy of other alternative opioids such as methadone and buprenorphine. While clinical studies had previously investigated the analgesic efficacy of methadone and buprenorphine, none had utilised a validated multidimensional pain assessment tool (PAT) designed to assess acute pain in cats, which called into question their findings. To address this need, the author conducted clinical studies in cats to evaluate the individual perioperative analgesic efficacy of each of methadone and buprenorphine compared to butorphanol. Surgical ovariohysterectomy (OVH) was used as a model of acute pain. Postoperative pain was assessed using a multidimensional PAT recently validated for assessment of acute pain in cats following OVH. Within the conditions of the studies documented within this thesis, methadone administered subcutaneously as a premedication at 0.6 mg/kg provided effective postoperative analgesia for at least 6 hours following OVH surgery in most cats. Buprenorphine administered intramuscularly as a premedication at 20 µg/kg with repeat administration of the same dose at the time of surgical wound closure, provided effective postoperative analgesia for at least 6 hours following OVH surgery. In contrast, intramuscular administration of a single premedication dose of 20 µg/kg IM buprenorphine prior to anaesthesia provided inadequate postoperative analgesia and additional analgesic drugs were required. All dosing regimens involving butorphanol resulted in unacceptable numbers of cats requiring rescue analgesia, suggesting that butorphanol is not a suitable choice of opioid for the provision of postoperative analgesia in the cat at these doses. An additional clinical study was conducted to examine the effect of atipamezole on analgesic requirement, and hence its potential confounding influence on the findings reported. Results from this follow up study suggested that the administration of atipamezole at a clinically recommended dose does not significantly affect the immediate postoperative pain scores in cats following OVH. Physiological data collected during the anesthetic period of these studies indicated that none of the opioids administered caused severe deleterious effects on vital functions and that all could be used safely for premedication prior to general anesthesia. In part dependent upon the findings from these studies, methadone and buprenorphine received registration approval with the Australian Pesticides and Veterinary Medicines Authority (APVMA) for use in cats within Australia. This will in part assist in addressing the undertreating of feline patients experiencing acute perioperative pain by providing clinicians with more efficacious registered drug options. In the course of evaluating the findings of the clinical studies reported in this thesis within the context of the wider peer-reviewed literature, it became apparent that inconsistencies existed in the way analgesic efficacy of opioids in cats is reported. Many of these inconsistencies arise due to variations in the quality and rigour with which a particular study is designed (including the use of non-validated PATs) and reported. To address these inconsistencies, a critical appraisal tool was developed for reviewing analgesia studies (CATRAS) involving animals and human patients who are incapable of self-reporting pain. Content validation was achieved using Delphi methodology via panel consensus. A panel of six experts reviewed the CATRAS in three rounds and quantitatively rated the relevance of the instrument and each of its quality items to their respective domains. The resulting 67-item critical appraisal tool will now provide a means of critically and quantitatively assessing the quality of analgesia trials involving subjects incapable of self-reporting pain for use in systematic reviews and meta-analysis studies.

Complication rates following emergency laparotomy surgery are high, with organ dysfunction being a commonly encountered post-operative complication. Given the endothelium acts as the interface between the systemic circulation and the organs, its function is vital to maintaining organ health. The endothelium is in a constant state of flux, impacted largely by the local environment of which it is a part. In the presence of wide-spread systemic inflammation, inflammatory mediators precipitate change to the structure of the endothelial glycocalyx. These changes result in shedding of the endothelial glycocalyx and alteration of the endothelial phenotype. The endothelium may, as a result, lose the capacity to regulate vasomotor tone, and shift toward a pro-inflammatory and pro-coagulant state. This predisposes to reduced tissue oxygen delivery, and organ dysfunction may ensue. This thesis aimed to answer two key questions: does surgical trauma induced in canine patients undergoing emergent abdominal surgery invoke a systemic inflammatory response and subsequent endothelial activation? And if so, does lidocaine, a proposed immunomodulatory drug, mitigate this effect when given in the post-operative period? Chapter two provides a detailed review of endothelial structure and function, and current literature pertaining to systemic inflammation and endothelial activation in the context of abdominal surgery. Chapter two also examines the literature regarding the proposed mechanisms through which lidocaine acts as an immunomodulatory drug, and reviews publications that investigate the use of lidocaine as an anti-inflammatory drug in human patients after abdominal surgery. Chapter three is a randomized, blinded clinical trial quantifying the effect of emergency abdominal surgery on the concentration of markers of systemic inflammation and endothelial perturbation in canine patients in the post-operative period. The trial also assessed the potential use of lidocaine as a post-operative immunomodulatory therapy in dogs having undergone laparotomy. Fifty canine patients undergoing abdominal surgery were enrolled in the study. Patients were randomized into two separate groups: a study group receiving lidocaine intravenously, and a control group receiving 0.9% NaCl intravenously for a twelve-hour period following abdominal surgery. Blood samples were gathered prior to surgery, followed by six and twelve hours post-operatively. Concentrations of markers of systemic inflammation (IL-6) and markers of endothelial perturbation (VEGF and HA) were quantified via means of ELISA at each time point. Results revealed a significant increase in the concentration of markers of systemic inflammation and endothelial perturbation in post-operative blood samples. No immunomodulatory or endothelial preserving effect of lidocaine was appreciated.

The mechanism of placental detachment in the mare has not been investigated in detail and as such remains poorly understood. Research in other species provides strong evidence that parturition and fetal membrane detachment from the endometrium requires involvement of the immune response. Research in horses has also suggested that movement of leukocytes in the umbilical vessels, and their attraction to fetal membranes at parturition, contribute to placental detachment. The aims of this study were to 1) confirm whether a leukocyte “concentration gradient” exists between the equine neonatal umbilical artery and vein, and whether this gradient was associated with fetal membrane retention time 2) to investigate the expression of pro-inflammatory cytokines in the equine endometrium and fetal membranes at parturition using quantitative real time polymerase chain reaction (RT-qPCR) and 3) to localise pro-inflammatory cytokine proteins within the equine endometrial and fetal membrane tissues using immunohistochemistry to correlate with gene expression. Data was collected from 33 spontaneously foaling mares and their foals on a single Thoroughbred stud farm in New Zealand. Umbilical artery and vein blood, fetal membrane and endometrial biopsy tissue samples were obtained. The average fetal membrane retention time was 92 (+/-130) minutes. The median fatal membrane retention time was 47 minutes (IQR 30). There was a significantly higher number of total leukocytes, lymphocytes and neutrophils in the equine neonatal umbilical artery than the umbilical vein at parturition (p <0.001). Fetal membrane retention time tended to decrease with an increase in umbilical leukocyte concentration gradient but did not reach statistical significance (p = 0.098). The mRNA expression of IL-1, IL-6 and IL-8 was significantly greater in the endometrium of foaling mares compared to control mares (p < 0.001). The mRNA expression of IL-8 was significantly higher in the fetal membranes of foaling compared to control mares (p < 0.001) whereas mRNA expression of IL-6 in fetal membranes was significantly lower between the two groups (p< 0.001). Whilst mRNA expression for IL-1 was lower in the fetal membranes of foaling mares compared to controls, the difference was not significant (p = 0.16). This study has demonstrated the presence of an innate inflammatory immune response in the placenta of spontaneously foaling mares, suggesting the role of inflammation in the detachment of the fetal membranes from the endometrium. The results from this study have highlighted the need to further investigate the mechanism of placental detachment in the mare and also understand the pathophysiology behind abnormal fetal membrane retention.

The eastern barred bandicoot, Perameles gunnii, is currently considered extinct in the wild, and is extant only as intensively managed reintroduced populations. Phillip Island, in Westernport Bay, Victoria, has been proposed as a potential reintroduction site for this species. Feral cats, Felis catus, the definitive host for the coccidian parasite Toxoplasma gondii, are present on the island, and eastern barred bandicoots are known to be susceptible to toxoplasmosis. The aim of this study was to investigate the epidemiology of T. gondii on Phillip Island, and the potential risk to the establishment of eastern barred bandicoot populations. The prevalence of T. gondii in the feral cat population on Phillip Island was determined using real time PCR (qPCR), and seroprevalence was determined using the modified agglutination test (MAT). A total of 161 feral cats was sampled between June 2016 and November 2017. Overall prevalence by qPCR was 79.5 % (95 % confidence interval [95% CI] 72.6 - 85.0), and overall seroprevalence was 91.8 % (95% CI 84.6 – 95.8). Additionally, the diagnostic sensitivity and specificity of both testing methods, and the true prevalence of T. gondii in feral cats on Phillip Island, were evaluated using Bayesian modelling. The sensitivity and specificity of the MAT in cats were estimated as 96.1 % (95 % credible interval [95% CrI] 91.5 – 98.8) and 82.0 % (95% CrI 65.1 – 93.3), respectively. The sensitivity and specificity of the qPCR method in cats were estimated as 90.1 % (95% CrI 83.5 – 95.6) and 96.2 % (95% CrI 82.2 – 99.8), respectively. The true prevalence of T. gondii infection in feral cats on Phillip Island was estimated as 90.2 % (95% CrI 83.1 – 95.2). Environmental contamination with T. gondii oocysts was assessed. Soil was collected from two sites on the island, Summerland Peninsula and Cape Woolamai, in September and October 2017. A total of 412 soil samples, from 206 sites, was tested. Soil samples were processed to isolate any protozoan organisms present and qPCR was conducted to specifically detect T. gondii DNA. Toxoplasma gondii oocysts were not detected by qPCR in any of the soil samples collected. Additionally, European rabbits, Oryctolagus cuniculus, were used as an indicator species to predict environmental exposure in eastern barred bandicoots. A total of 134 feral rabbits was sampled from August 2016 to November 2017. Overall prevalence of T. gondii in feral rabbits, as determined by qPCR on tissue, was 10.5 % (95% CI 6.3 - 16.8). Toxoplasma gondii infection was not identified in any rabbits collected from the proposed release site for eastern barred bandicoots, the Summerland Peninsula. Sixty-seven eastern barred bandicoots were released onto the Summerland Peninsula in October and November 2017 as part of an assisted colonisation trial. Sixty-one of these bandicoots were tested for prior exposure to T. gondii using the MAT. No animals were seropositive for T. gondii prior to release on to Phillip Island, and the population could be concluded to be free from disease, at a design prevalence of 6 %. Bandicoots were trapped for blood sample collection in November 2017 after approximately three weeks of habitation on the island (n = 23), and again in February 2018 after approximately four months (n = 21). No bandicoots demonstrated seroconversion to T. gondii on the MAT in either November or February. In November, this was sufficient to conclude that the population was free from infection with T. gondii, at a design prevalence of 6 %. However, in February, the sample size was too small to make this conclusion. Mortality data on the bandicoots was limited, due to difficulties associated with placing radio-transmitters on this species. However, one deceased animal was recovered. This animal had no evidence of T. gondii infection based on qPCR on tissues. These findings suggest that while prevalence of T. gondii infection is very high in feral cats on Phillip Island, other factors, such as low cat density at the site, may translate to a low toxoplasmosis risk for eastern barred bandicoots released onto the Summerland Peninsula. Further studies are required to understand the factors contributing to the observed low level of environmental contamination with oocysts. Additionally, future studies should focus on increasing the sensitivity of mortality surveillance in the eastern barred bandicoot population on the Summerland Peninsula. This will enable the collection of more robust mortality data, further elucidating the importance of toxoplasmosis in this population.

Real-time continuous glucose monitoring systems (CGMS) measure interstitial glucose concentrations, and have been used in the management of diabetes mellitus in people, dogs and cats. The devices are used for up to 72 hours, and provide glucose measurements every 5 minutes, with 288 data points provided in a 24-hour period. This provision of a detailed insight into glycaemic control over a longer period of time than traditional methods of monitoring holds the potential for improved management of diabetes mellitus. The primary aim of this study was to determine if CGMS (using the Guardian™ system) resulted in different clinical decision making compared with monitoring serial blood glucose curves and serum fructosamine concentration in diabetic dogs and cats. Secondary aims were to determine the incidence of nocturnal hypoglycaemia and rebound hyperglycaemia in diabetic dogs and cats. Continuous glucose monitoring and fructosamine measurement were performed in client-owned dogs and cats, both newly and previously diagnosed with diabetes mellitus. A retrospective serial glucose curve was plotted with glucose measurements every 2 hours obtained from the CGMs data. Results of the three monitoring modalities along with historical data (i.e. appetite, thirst, insulin dosage) were collated and a blinded review performed by two board certified small animal internal medicine clinicians. Statistical analysis showed a difference in clinical treatment recommendations for the management of diabetic dogs and cats when using CGMs versus both serial glucose curves and serum fructosamine. Nocturnal hypoglycaemia was seen in 14.6% of diabetic dogs and cats and the 9.8% had episodes of the Somogyi effect.

Designing and delivering effective, useful livestock health surveillance is a challenge for many countries. The observations of people in frequent contact with livestock, captured through passive surveillance, play an important role in many national surveillance systems. In Australia, the effectiveness of passive surveillance on sheep and beef farms has been limited by infrequent veterinary contact. Farm workers frequently observe signs of disease in livestock, but these observations are not captured by existing surveillance systems. This thesis therefore posed the question: can farmers’ observations be collected to generate useful surveillance information? Syndromic surveillance of farmers’ observations is one approach to increase data capture from extensive livestock farms. Chapter 3 describes the operation of a syndromic surveillance system collecting farmers’ observations of livestock health in Victoria, Australia, over its first two years of operation from 2014 to 2016. Survival analysis and classification and regression tree analysis were used to identify farm level factors associated with reliable participation, to inform future recruitment aimed at farmers who were willing and able to provide regular, timely reports. Farmers keeping only sheep were the most reliable and timely respondents, while farmers aged under 43 years or working full time on-farm had lower response rates than older farmers or part-time farmers. This chapter demonstrates that recording farmers’ observations of signs of disease using syndromes is a feasible and effective method to gather disease occurrence data. The utility of syndromic data is further investigated in Chapter 4, using the observations collected by the surveillance system to quantify ewe mortality on sheep farms in southern Australia. Ewe deaths were reported in 540 of 612 reports, describing 2106 individual deaths, with a median of 4 deaths per positive monthly report. Median mortality rates ranged between individual farms from 1 to 5 deaths/1000 ewes/month. The incidence rate ratio of mortality in the five months preceding and following lambing was 2.8 (95% CI 2.0 to 4.1) compared to the remaining seven months of the year. Overall ewe mortality could therefore be reduced through strategies targeted to improving peri-parturient ewe survival. In a subset of reports where veterinary contact was recorded, just 15% of reported deaths involved a veterinarian. Further investigation of how and why farmers respond to ewe deaths without veterinary support is needed, to determine the best farm management strategies to reduce mortality. Chapter 5 investigates Australian sheep farmers’ low rates of veterinary contact. The study aimed to understand why Australian sheep farmers chose not to contact veterinarians when their animals showed signs of disease, and what alternative approaches they took to managing unwell animals. Data were collected during three focus group discussions with sheep farmers in Victoria, Australia. Transcripts of those discussions were analysed using a modified grounded theory approach to develop a preliminary theory of Australian sheep farmers’ disease response behaviour. Critical steps in the decision-making process included the farmer recognising that action is needed, and then deciding what that action would be. The farmers reported having to decide whether they would act independently based on their previously experiences, or alternatively to seek advice. Veterinarians played a small but important role as potential advisors, alongside others including trusted farming friends and farmer discussion groups. Self-reliance and confidence in their knowledge and skills was highlighted as the main reason the farmers often chose not to seek veterinary advice. Rather than being seen as a barrier to effective passive surveillance, the actions that arise from farmers’ self-reliance when facing disease should be taken into account when designing novel surveillance approaches. A final consideration for observational disease surveillance is the selection of individuals to contribute data to the system. While characteristics associated with participation may guide recruitment as described in Chapter 3, it is also useful to target surveillance to farms that have increased risk of acquiring or disseminating disease. The movement of animals between farms contributes to infectious disease spread, and can be investigated through network analysis methods. Australia’s National Livestock Identification Scheme sheep movement records are suitable for such analyses, but are known to be a targeted subset of all sheep movement in the country. However, knowledge of the effect of sampling or incomplete network data on these studies is limited. In Chapter 6, a simulation algorithm is presented that provides an estimate of required sampling proportions based on predicted network size, density and degree value distribution. The algorithm may be applied a priori to ensure network analyses based on sampled or incomplete data provide population estimates of known precision. Results demonstrate that, for network degree metrics, sample size requirements vary with sampling method. Where simulated networks can be constructed to closely mimic the true network in a target population, this algorithm provides a straightforward approach to determining sample size under a given sampling procedure for a network metric of interest. Chapter 7 then presents analysis of National Livestock Identification Scheme sheep movement data for Victoria, Australia. The sheep movement network in Victoria shows typical livestock movement network characteristics including scale-free and small-world topology, small diameter and short average path lengths, supporting the assumption that disease could spread rapidly in the state through sheep movements if it were not detected rapidly. Victoria’s position as a net importer of sheep and sheep flow is confirmed, driven substantially by the activity of saleyards (livestock markets) and abattoirs. Little variation within or between years in overall movement patterns were detected. While most farms are connected to a very small number of properties in the network, small subsets of farms demonstrate high degree values (being directly connected to many other properties through incoming out outgoing animal movements) or high frequency of sheep purchases or sales. These farms may be useful targets for emerging surveillance methods that can be implemented on-farm. Together, these studies provide new information about the Australian sheep industry and the feasibility of new surveillance approaches to improve the effectiveness of surveillance. By describing farmer behaviour, livestock movements patterns and the feasibility of syndromic surveillance approaches to capture farmers’ observations of signs of disease, these studies justify further development and implementation of novel surveillance approaches in Australia and serve as an example for other countries facing similar surveillance challenges. While there is no ideal surveillance system, integrating new approaches into wider surveillance strategies can improve the quality of information generated by surveillance, to better describe true disease states in the population and drive appropriate response activities.

Improving wildlife population health requires an understanding of the infectious agents within those populations. Historical accounts of herpesviruses in marsupials indicate that they can have a significant impact on animal health. This evidence is strongest for the macropodid alphaherpesviruses, but with improvements in molecular diagnostics the discovery of novel viruses has outpaced our understanding of their impact and significance. This thesis aimed to expand our knowledge of marsupial herpesviruses by examining relationships between marsupial herpesviruses and other herpesviruses, and by describing the clinical significance of infection. This thesis also aimed to improve diagnostic tools for detecting herpesvirus infection in marsupials. The core genomes of three marsupial herpesviruses were determined; macropodid alphaherpesvirus 1 (MaHV1, infecting wallabies), phascolarctid gammaherpesvirus 1 (PhaHV1, infecting koalas) and vombatid gammaherpesvirus 1 (VoHV1, infecting wombats). MaHV1 had a similar genome arrangement to other simplexviruses, but contained gene clusters that may be unique to the macropodid simplexviruses. PhaHV1 and VoHV1 had a shared gene arrangement and were likely to have speciated from a common ancestor. Over 30 new ORFs were identified within the genomes. Functional enzymatic characterisation was performed on two viral NTPDase homologs encoded within the two gammaherpesviruses. NTPDase activity was confirmed for the PhaHV1 homolog but not the VoHV1 homolog. Koalas are host to two divergent gammaherpesvirus species, PhaHV1 and -2. To understand the clinical significance of each individual virus a large molecular epidemiological study of 810 koalas from 7 separate geographic regions was conducted. Samples were tested using a rapid and differential PCR-HRM assay. Available signalment and clinical observation data was analysed in comparison to infection status through univariable and multivariable logistic regression analysis. Additional factors considered were location, year, body condition, fecundity in females, as well as the presence of other infectious agents (Chlamydia pecorum and koala retrovirus). PhaHV1 and -2 were present in 17% and 22% of koalas tested (state-wide), although some variation from the state average was observed in particular populations. Neither virus was associated with a particular sex. PhaHV1 detection was uniquely associated with the presence of koala retrovirus as well as increasing age. PhaHV2 detection did not change with age, which may indicate differences in how these two viruses are acquired over the life of the animal. Both viruses were positively associated with genital tract abnormalities, lowered fertility in females, emaciated body condition, urinary tract infection (wet bottom) and detection of C. pecorum, although the strength of these associations varied by sex and herpesvirus species. To further the development of herpesviruses serological tools, this thesis examined the ability of four commercially-available immunoglobulin-binding reagents to bind serum antibodies from 17 species within the Marsupialia and Monotremata. Serum samples were assessed for binding using immunoblots and ELISAs to three microbially-derived proteins; staphylococcal protein A, streptococcal protein G and peptostreptococcal protein L, and to an anti-kangaroo antibody. The inter- and intra-familial binding patterns of the reagents to serum immunoglobulins varied and evolutionary distance between animal species was not an accurate predictor of the ability of reagents to bind immunoglobulins.

Intra-mammary antibiotics are widely used, and are an important contributor to the control of bovine mastitis. The concentrations of antibiotic drugs in milk after intra-mammary administration is highly important for food protection and treatment efficacy. In this project, novel mathematical and statistical models were developed to describe the pharmacokinetics of antibiotic drugs after intra-mammary administration. This approach combined physiologically-based pharmacokinetic models to capture the unique features of the mammary gland, and large-scale mixed-effects analysis to understand these effects at the level of the treated population. The developed models perform well across various drug types and production conditions, and it is proposed that these models are suitable as a generalized method for the description and analysis of drug concentration in milk. Further, the developed methods are applied to study the statistical performance of the various existing methods used to study milk residues in the regulatory context. A derived model is discussed, which is suitable for application to regulatory conditions for the determination of milk discard times.