University of Thessaly (UTH)
A study of ecosystemic interactions in small ruminant farms within the context of one health and precision medicine: environmental variables, climatic influence, health problems and diseases, quality of milk, interactions with domesticated and wild animals, development of prediction models
Abstract
dc:descriptionThe present thesis focuses in describing and characterizing small ruminant farms in Greece, through a viewpoint that encompasses and considers the entire ecosystem of these farms. This includes the livestock on farm, i.e., sheep or goats, other animals present on there (domestic (e.g., other livestock, farm dogs) or not (e.g., rodents)), also sheep or goats from nearby farms and wildlife near the farm, people associated with the farm (e.g., farmers, work staff, visitors), other non-living environment (e.g., infrastructure or grazing areas) and, of course, microorganisms (pathogenic or non pathogenic) present on the farm, as well as the climatological conditions that prevail at the location of the farm. Specific objectives of the thesis are as below.(a) The presentation of changes in climate-related parameters from 1989 to 2019, at 444 locations throughout Greece, where small ruminant farms have been based, and the assessment of associations of the changes in these parameters with data related to small ruminant health,(b) The assessment of beliefs and thoughts of small ruminant farmers in Greece on climate changes and potential effects on animal farming by carrying out a questionnaire-based study,(c) The exploration of associations between climatic variables and the presence of health problems in sheep and goat farms and the comparison of these to the health management practices applied in the farms,(d) The survey of the presence of respiratory infections in lambs and kids and the identification of variables associated with the presence of these infections in the farms, with emphasis on the potential role of industrial sites near farms,(e)The evaluation of associations of reproductive performance in sheep and goats farms with climate conditions at farm location,(f) The evaluation of associations between climate-related variables and parameters characterising the quality of bulk-tank milk produced in sheep and goat farms in Greece and the evaluation of associations between climate-related variables and the recovery of pathogens from the bulk-tank milk produced in sheep and goat farms,(g) The evaluation of associations between climate-related variables and the recovery of pathogens from the teatcups of milking parlours in sheep and goat farms in a field study, the assessment of the growth and spread of Staphylococcus epidermidis, a confirmed mammary pathogen, under in vitro conditions, on teatcups used in milking parlours and the potential effects of environmental conditions,(h) The evaluation of the importance of climate-related variables in the vaccination patterns applied in sheep and goat farms in Greece and the assessment of potential interactions between these factors and previously established important health management- and human resources-related factors applied in the farms,(i) The description of the presence of animals other than sheep or goats in small ruminant farms and of their characteristics,(j) The investigation of the gastrointestinal nematode infections in dogs in small ruminant farms in Greece, the elucidation of potential predictors for the infections and the description of practices related to the administration of anthelmintics to dogs,(k) The evaluation of occurrence of wild mammals near sheep and goat farms, the quantification of adverse outcomes including livestock losses due to predation, parasite presence and load, rates of abortion, neonatal diarrhoea and lameness, the evaluation of the association of wild mammal co-occurrence (e.g., predation) and co-grazing with wild ruminants (e.g., as reservoirs for parasites) in influencing livestock health throughout the national Greek flock and herd and the evaluation of relevant associations with somatic cell counts in the bulk-tank milk of the farms,(l) The description of the presence of hunters near the location of sheep and goat farms,(m) The identification of interactions between all the variables, climate-related and non climate-related, and of their potential effects on parameters characterising the quality of bulk-tank milk produced in sheep and goat farms in Greece and on the recovery of pathogens from that bulk-tank milk,(n) The development of a computational model with which predictions regarding the future incidence of abortion and lameness in small ruminant farms could be performed, by taking into account potential changes in climate-related variables and(o) The development of a computational model with which predictions regarding the presence of genetic material of Coxiella burnetii in the bulk-tank milk of small ruminant farms, by taking into account management-related and climate-related variables. The thesis is divided into 17 chapters followed by the General Discussion. Chapter I appraises in brief the review of the relevant literature. The Chapter is subdivided into four Parts, which review the role of the changes of the temperature (Part B), the role of lack of water (Part C), the role of climate change and animal health (Part D) and the role of natural disasters (Part D), as exemplified for small ruminant health, welfare and production. Chapter II outlines the general procedures applied in the study, specifically, the source of the data employed in this work (Part A) and the derivation of data regarding climate variables at the locations of the farms (Part B). Data regarding 325 sheep farms and 119 goat farms were employed during this work. Chapter III presents the climatic variables from 1989 to 2019 were obtained for 444 locations with small ruminant farms throughout Greece. During this period, significant increases were noted in temperature-related parameters (annually 0.05 °C for average temperature and 0.14 °C for temperature range) and precipitation. There were significant differences in climatic conditions be-tween locations of farms in accord with the management system applied therein, as well as in accord with the breed of animals on the farms (e.g., higher average temperature in locations with Greek breeds, higher temperature range in locations with imported breeds). There were significant associations of temperature-related parameters with the annual frequency of cases of neonatal hypothermia seen at a veterinary hospital, as well as with the average proportion of Haemonchus contortus larvae in faecal samples and the frequency of cases of H. contortus resistance reported by a veterinary parasitology laboratory. Chapter IV presents the beleifs and thoughts of small ruminant farmers regarding climate changes and potential effects on animal farming, by means of a questionnaire with 21 questions. A total of 96 answers were received, mainly (78%) from sheep farmers. Most farmers (93%) believed that changes in climate had occurred during the last 20 years. These referred mostly to changes in temperature and precipitation. Most farmers (60%) reported the occurrence of extreme weather conditions at the locations of their farms. Among them, the majority (55%) reported damages to their farm due to these conditions, but fewer (28%) reported the death of animals as a consequence. Most had heard previously the term ‘climate change’. Their majority reported that they considered this as a general threat for people and even a potential significant problem for animal production and expressed their intentions to make appropriate changes in their farms to address its effects. Chapter V explores associations between climate conditions and health problems in sheep and goat farms in Greece. In Part B, data regarding the annual incidence rate of abortion, clinical mastitis, lamb pneumonia and lamb diarrhoea are presented. The annual incidence rate for abortion was 2%, for clinical mastitis 4%, for lamb pneumonia 1% and for lamb diarrhoea 8%. In multivariable analyses, climatic variables emerged as significant predictors for abortion, specifically high annual precipitation at the farm location (p = 0.024), and for lamb diarrhoea, specifically high average annual temperature range at the farm location (p < 0.0001), but not for clinical mastitis or lamb pneumonia. The potential effects of climatic variables were found to be more important in lambs than in adult animals. In Part C, data regarding the assessment of animals on farms for the presence of lameness are presented. The within farm prevalence rate of lameness varied from 0% to 25% in sheep flocks and from 0% to 30% in goat herds. The mean (median) within farm prevalence rate among sheep farms was 2% (0%); among goat farms, it was 3% (0%). Multivariable analysis for within farm prevalence of lameness revealed three significant predictors in sheep farms: application of vaccination against foot-rot (p = 0.0002), increased precipitation at the farm location (p = 0.002) and longer annual grazing period (p = 0.006), and one in goat farms: increased precipitation at the farm location (p < 0.0001).In Part D, data for parasitic burdens in animals of the farms, obtained by carrying out established parasitological techniques in faecal samples collected in the farms, are presented. The multivariable analysis revealed associations of the following climatic variables with the epg counts: relative humidity for 1 year prior to sampling (p < 0.0001), relative humidity for 6 months prior to sampling (p = 0.0006), Earth surface temperature for 6 months prior to sampling (p = 0.003), Earth surface temperature for 1 year prior to sampling (p = 0.013) and average precipitation for 1 year prior to sampling (p = 0.015). Associations of climate-related variables with specific parasites were seen for the proportion of Haemonchus contortus and for the presence of Nematodirus spp. eggs and Eimeria spp. and Cryptosporidium spp. oocysts in samples. Chapter VI explores associations between the farm environment and the incidence of pneumonia in lambs and kids. In multivariable analysis, the proximity of farms to industrial sites notably emerged as significant predictor for the incidence of pneumonia of lambs and kids in both sheep and goat farms (p < 0.015).Chapter VII presents associations of climate-related variables with outcomes related to reproductive performance only in farms, in which reproductive control was not used: significant associations were seen for the ‘Length of matings’ with the average maximum temperature at 2 m for 1 year prior to the visit (p = 0.027), for ‘Total lambs / kids born per ewe / doe’ with the average mean temperature at 2 m for 1 year prior to the visit (p = 0.048) and for the ‘Age of culling females’ with the average maximum temperature at 2 m for 1 year prior to the visit (p = 0.003). In contrast, in farms, in which reproductive control was used, no such associations were found. Chapter VIII explores associations between climate-related factors and parameters related to quality of the bulk-tank milk produced in sheep and goat farms. In Part B, data for fat content, protein content, somatic cell counts and total bacterial counts of bulk-tank milk are presented. In Part C, data for the presence of Staphylococcus aureus, biofilm-forming staphylococci, antibiotic-resistant (to at least one antibiotic) staphylococci, Listeria spp. and C. burnetii of bulk-tank milk are presented. Chapter IX describes associations between climate-related factors and the presence of staphylococci on teatcups through field, computational and experimental studies. In Part B, the results of a large field study in sheep and goat farms and data for the association of the isolation of staphylococci from teatcups of milking parlours with the average precipitation for 7 days prior to sampling (p = 0.003) are presented. In Part C, the results of a computational study to assess strains for Staphylococcus epidermidis for subsequent experimental work are presented. The MLST database included 1593 isolates from 46 countries: 1215 of human, 195 of environmental and 134 of animal origin, distributed into 949 sequence types (STs) and cumulatively with 450 alleles therein. Among mastitis isolates, bovine isolates were distributed into 36 different STs and ovine ones into 15 STs. The 33 isolates from ovine mastitis in Greece were in 15 different STs, 6 of these (ST677, ST678, ST700, ST 709, ST710, ST711) assigned for the first time; in addition, 5 al-leles (65 for arcC, 59 for aroE, 56 and 57 for gtr and 48 for tpiA) were identified for the first time. The spanning tree of these isolates included 15 nodes and 14 edges (i.e., branches). Among these isolates, 19 showed resistance to antimicrobial agents (tetracycline, penicillin, fucidic adic, erythromycin, clindamycin, cefoxitin). Resistance-related genes (tetK, tetT, msrA, tetM, tetS, ermC, mecA) were detected. There was no association between STs and resistance to antimicrobial agents. Isolates with antimicrobial resistance were recovered more often from flocks where hand-milking was practised. In Part D, the results of two experiments to evaluate the growth of S. epidermidis on teatcup material are presented. In the first experiment, the growth of two S. epidermidis isolates (one biofilm-forming and one not) on teatcups for cattle (made of rubber) or sheep (made of silicone) were assessed in nine multiplicates for 24 h post-smearing on the teatcup surface. Staphylococci were smeared on an area of 0.0003142 m2 on the material and their growth and expansion further on were monitored for 24 h. There were no differences in the frequency of recoveries between the two isolates (p > 0.82 for all comparisons). There were more recoveries from sheep teatcups than from cattle teatcups: 1280/1728 (74%) versus 942/1728 (55%), for both isolates (p < 0.0001). Significance was observed only 6 h to 15 h after smearing (p < 0.0001 for all comparisons). The median speed of linear dissemination of the isolates was 0.00000021 m s-1 on cattle teatcups and 0.00000033 m s-1 on sheep teatcups (p < 0.0001). In the second experiment, the growth of the isolates on new or used teatcups made of silicone for use in milking parlours for sheep, was assessed for 24 h after the application by smearing on the surface of the teatcup and monitored for 24 h at varying ambient conditions: temperature 21 °C or 31 °C and humidity 60% or 80%. No differences were evident between the two isolates in the frequency of recoveries in any of the conditions tested (p > 0.75 for all comparisons). Recovery rates were higher in humidity 80% compared to humidity 60%: 1678/2016 (83%) versus 1282/2016 (64%) (p < 0.0001), and in temperature 31 °C compared to temperature 21 °C: 1525/2016 (76%) versus 1435/2016 (71%) (p = 0.001). Recovery rates were also higher from new teatcups compared to used ones only in humidity 60%: 744/1008 (74%) versus 538/1008 (53%) (p < 0.0001). Humidity 80% was associated with higher speed of linear dissemination of the isolates on teatcup surface compared to humidity 60%: 0.000000640 (6.40×10–7) m s−1 versus 0.000000322 (3.22×10–7) m s−1 (+99%) (p < 0.0001); no such association was seen with higher temperature: 0.000000509 (5.09×10–7) m s−1 versus 0.000000453 (4.53×10–7) m s−1 for temperature 31 °C and 21 °C (+12%) (p = 0.29).Chapter X explores the importance of climate-related variables in the vaccination patterns applied in sheep and goat farms in Greece and assesses potential interactions between these factors and previously established important health management- and human resources-related factors applied in the farms. Vaccination patterns against chlamydial abortion, clostridial infections, contagious agalactia, contagious ecthyma, foot-rot, paratuberculosis, pneumonia or staphylococcal mastitis were assessed. Climatic variables (2010–2019 and 2018–2019) were used. Climatic variables had a higher association with vaccinations against infections in sheep flocks (26 associations) than in goat herds (9 associations) (p = 0.002) and in farms with semi-extensive or extensive management (32 associations) than in farms with intensive or semi-intensive management (8 associations) (p < 0.0001). In 26 cases (39% of all analyses evaluated), the climatic variables overshadowed the management- and human resources-related variables assessed as significant predictors for vaccination. In most cases, these referred to sheep flocks (nine cases) and farms with semi-extensive or extensive management (eight cases). For all eight infections, there were changes in the climatic variables found to be significant predictors from the 10-year dataset to the 2-year dataset. Chapter XI describes other domestic animals, apart from sheep or goats, present in small ruminant farms. Such animal species, apart from sheep or goats, were present in 98% of farms, more frequently in farms under intensive or semi-intensive (99%) than under semi-extensive or extensive (96%) management system (p = 0.033). Dogs, cats and poultry were the other domestic animals present more frequently on the farms. When considering sheep or goats as one (1) ‘animal unit’, the median number of ‘animal units’ of other domestic animals on the farms was 4.5 (14.7). There was a clear association between the number of sheep / goats on the farms and the ‘animal units’ related to the other domestic animals on the farms (p = 0.003)Chapter XII deals with an investigation of gastrointestinal nematode infections in dogs in the fsmall ruminant farms, the elucidation of potential predictors for these infections and the description of practices related to administration of anthelmintics to dogs. Faecal samples were collected from the dogs in the farms and processed by means of conventional parasitological techniques, specifically, a combined sedimentation flotation technique. There were dogs in 93% of the farms, with a median number of four dogs per farm. The following variables were associated with the presence of dogs in the farms: the presence of wild mammal predators near the farms, the increased daily period of farmers’ presence at the farm, goats as the livestock species at the farm and the management system applied in the farm. Helminth eggs were detected in samples from 73% of the farms. The main helminth eggs detected were those of hookworms (Uncinaria/Ancylostoma) and Toxocara canis, in 69% and 51% of the farms, respectively. In our multivariable analyses, an association emerged between the presence of canid predators near a farm and the detection of these helminths in faecal samples: in 76% and 60% of the samples, respectively, versus in 58% and 39% of the samples from farms with no canid presence. Of farmers with dogs, 16% reported that they omitted the administration of anthelmintics to the animals. In multivariable analysis, the semi-extensive or extensive management system applied in the farm, the lower annual milk production per animal and the lack of collaboration with a veterinary practice were the significant predictors for the omission of anthelmintic administration to the farm dogs. There was also a clear association in the omission of anthelmintic administration to the dogs and to the livestock on the farm. The most frequently administered anthelmintic was praziquantel, which was used in 94% of the farms. Chapter XIII discusses the presence of wildlife mammals near sheep or goat farms and potential associations with health problems of livestock. In Part B, data for the presence of wild mammals near small ruminant farms and their association with sheep and goat health-related parameters, specifically, parasite burden, abortions and neonatal diarrhoea are presented. Wild mammals were present near 92% of farms; these included fox (74%), wild boar (51%), grey wolf (40%) and golden jackal (23%) mainly. Median wild mammal species richness was 2 wildlife species near sheep and 3 wildlife species near goat farms, and was positively correlated with altitude. Wild mammals near farms were associated with warmer areas with greater seasonality. Livestock losses to wild mammals occurred at 29% of sheep and 56% of goat farms and more frequently in those under semi-extensive or extensive management system; it included predation by grey wolves (at 60% of farms) and golden jackals (at 41% of farms). Sheep and goats that co-grazed with wildlife ruminants had higher faecal parasite egg counts (348 epg), higher Haemonchus contortus proportion in faeces (33%), higher abortion rate (2%) and higher neonatal diarrhoea incidence (9%) than at farms where no such co-grazing occurred (231 epg, 29%, 0%, 2%, respectively).In Part C, we describe an investigation between the association between wild canid predators near sheep farms throughout Greece and somatic cell counts in bulk-tank milk as a reflection of milk quality. Sheep farms were divided into three groups: Cohort A (farms with no reports of wild canid predators nearby), B (farms with canid predators (golden jackal and grey wolf) nearby yet with no experience of livestock losses to predation) and C (farms with canid predators nearby and livestock losses to predation). Somatic cell counts in bulk-tank milk of Cohort C farms were significantly higher, +43% and +29%, compared to those for Cohorts A and B, respectively: 0.617×106 cells mL-1 versus 0.433×106 or 0.477×106 cells mL-1, respectively. The presence of wild canid predators near sheep farms was associated with lower quality milk potentially indicative of stress consistent with the potential effects of a landscape of fear. Chapter XIV describes the hunting activity near small ruminant farms. Hunters were reported to be present near 89% of the farms, more frequently near goat (95%) than sheep (87%) (p = 0.022) farms, near farms under semi-extensive or extensive (94%) than under intensive or semi-intensive (85%) management (p = 0.003) and near farms at higher altitude (177 m versus 98 m) (p = 0.029). In most cases (94%), presence of hunters in the vicinity of farms was reported to occur three days per week, with a median distance of 500 m from farms. In most cases (84%), the interest of hunters referred to European brown hares and / or Eurasian woodcocks. Chapter XV reports the identification of interactions between all the variables, climate-related and non climate-related, and of their potential effects on parameters characterising the quality of bulk-tank milk produced in sheep and goat farms and on the recovery of pathogens from that bulk-tank milk. In total, 16 climate-related and 5 non climate related variables emerged as predictors for the fat (n = 5 in sheep and n = 11 in goat farms) and protein (n = 6 in sheep and n = 4 in goat farms) (p < 0.04 for all comparisons) content in the bulk-tank milk. For increasing somatic cell counts, four variables emerged as significant predictors: the increasing wind speed for 7 d prior to sampling (p = 0.023), the use of hand-milking (p = 0.023), the month of the lactation period at sampling (p = 0.041) and the presence of wild mammal predators near the farm (p = 0.043). For the presence of biofilm-forming staphylococci in the bulk-tank milk, the temperature of the water used to clean the milking parlour emerged as predictor (p = 0.016) in farms in which machine-milking was applied, whilst the mean temperature at 2 m for 7 days prior to sampling emerged as predictor (p = 0.010) in farms in which hand-milking was applied. For the presence of antibiotic-resistant (to at least one antibiotic) staphylococci in the bulk-tank milk, the number of annual disinfections applied in the farms (p = 0.0008) and the month into lactation stage at sampling (p = 0.011) were the significant predictors. For the presence of Listeria spp. in the bulk-tank milk, the presence of pigs in sheep / goat farms (p < 0.0001), the number of animals in the farm (p = 0.0005) and the relative humidity for the 15 days prior to sampling (p = 0.029). For the presence of C. burnetii in the bulk-tank milk, the average wind speed at 10 m for 15 d prior to sampling (p < 0.005) and the application of intensive or semi-intensive management system in farms (p < 0.030).Chapter XVI presents the development of a computational model for predicting by means of machine learning tools the annual incidence rate of abortion and lameness in sheep and goat farms. Computational models were constructed for the prediction of these two target values, through the use of 6 and 13 independent variables, respectively, related to health management and climate-related factors prevalent in the farm. Five machine learning tools were employed: decision trees (49 different models evaluated), random forests (12 models), XGBoost (192 models), k-nearest neighbours (96 models) and neural networks (216 models) (in total, 1130 evaluations were performed). Long Short-Term Memory (LSTM) networks were employed, in order to predict values for climate-related variable that would prevail at the locations of the farms in the year 2030; these predicted values were employed in the final prediction modelling. Models and tools found with the lowest mean squared error (MSE) among and between the five tools used to test predictions for each target value were selected. The lowest mean squared error was obtained for the annual incidence rate of abortion with k-Nearest neighbours and Neural networks and for the annual incidence rate of lameness with Random forests and k-Nearest neighbours. For abortion, the median within farm annual incidence rate was predicted to increase from 0% in 2019 to 3% in 2030 (p < 0.0001); abortion cases were predicted to occur in all 444 farms in 2030 (p < 0.0001). For lameness, the median within farm annual incidence rate was predicted to increase from 0% in 2019 to 1% in 2030 (p < 0.0001); lameness cases were predicted to occur in all 444 farms in 2030 (p < 0.0001). By modifying the health management applied in the farms, the predicted annual incidence rate of abortion could be predicted to be significantly reduced to 2% (p < 0.0001); however, the annual incidence rate of lameness in the small ruminant farms in the study could not be reduced significantly and was predicted to be 1% (p > 0.05).Chapter XVII presents the development of a computational model for predicting by means of machine learning tools the presence of C. burnetii in the bulk-tank milk of sheep and goat farms. Computational models were constructed for the prediction of the target value, through the use of 10 independent variables related to health management and climate-related factors prevalent in the farm. Two machine learning tools were employed: XGBoost (81 models) and neural networks (100 models) (in total, 181 evaluations were performed). As data related to the target value were imbalanced, the model development data involved data-level solutions like oversampling through the use of Synthetic Minority Oversampling Technique (SMOTE), in order to create synthetic minority samples, undersampling to reduce the majority class or hybrid methods combining both. Tool found with the highest accuracy, precision, recall and F1 score was selected. The highest of these measures of quality were obtained with XGBoost. For the presence of C. burnetii in the bulk-tank milk of sheep and goat farms, accuracy, precisions, recall and F1 score were 97%. By applying analysis for SHAP values, it was found that the management system applied in farms and the wind speed for 15 d prior to the assessment were the variables that mostly influenced the prediction outcome. This thesis presents, for the first time, a systematic study of the sheep and goat farms in Greece through an ecosystemic point of view. This ecosystemic viewpoint into sheep and goat farms is a novel approach in itself. The study, in its entirety, also presents many novel facets of the interactions between livestock on the farms: sheep and goats, and the totality of their environment: climate, infrastructure, other domestic animals on the farms, wildlife, in the food-producing chain. In their ensemble, the findings can be useful tools in the health management of small ruminants, by providing evidence-based support, within the scope of a precision livestock medicine. The associations with some climatic factors found in selected pathological conditions indicate that, during complex pathophysiological processes, environmental factors may, from time to time, have some effect. However, other interactions (e.g., from other domestic animals or wildlife) also play a significant role in disease processes. However, it should be noted that the overall associations recorded in the present work may not always be only the direct effect of climatic factors on disease outbreaks and may not always reflect an association of the need to protect against infection outbreaks as climatic conditions change. It should be noted that these may also be the results of perceptions of farmers, as well as the consequences of an indirect influence of climatic factors on disease patterns (e.g., through the effects on tolerance and susceptibility of the various animal breeds to varying weather conditions).The findings are particularly useful within the context of precision livestock farming. Future studies can focus on formulating health management, in accordance with climate factors. Also, setting the optimum season(s) for applying various health management tasks in the farms can be based on the circulation of the pathogens, the risks for the development of diseases and the stage within the annual production cycle of the animals. In farms where climate conditions were found to have an impact on clinical problems, management practices should be applied to contribute to the efficient control of the respective clinical problems. The overall results of the study can be used to scientifically update the relevant knowledge about conditions in sheep and goat farms. Farming conditions should be modified as the effects of differing dynamics or interactions prevailing in there change, according to time of year, technological changes and pathogen evolution. The findings suggest that, potentially, relevant algorithms could be designed to take into account climate factors for the formulation of health management programs in small ruminant farms. Future studies may focus on how variations and changes in the environment of farms can be translated into on-farm metrics to understand the impacts on livestock health and welfare. More specific conclusions derived from the results of the present thesis are as follows.(a) Long-term changes in climate variables at the locations of sheep and goat farms have occurred and have been presented coupled with their potential effects on the health of these animals. Although climatic factors may have some effect in diseases, other interactions also play a significant role in disease processes. There is a need for further research into the response of animals to challenges from the environment.(b) In general, farmers are aware of the importance of climate-related factors in the management of their farms. Nevertheless, work should be performed to increase their awareness and the changes to be implemented to counteract potential future adverse effects of climate-related factors.(c) The study provided, for the first time in Greece, extensive evidence regarding the importance of climatic conditions for the occurrence of diseases of sheep and goats. For abortion, lameness and lamb diarrhoea, a significant impact of climate conditions was found. In contrast, no such impact was seen for clinical mastitis and lamb pneumonia. Precipitation and temperature were recognised as the most important climate-related variables. The potential effects of climatic conditions were more important in younger animals. Management practices should be applied to contribute to the efficient control of the respective clinical problems. Specifically with regard to anthelmintic treatments, these can be scheduled taking into account climatic effects, in order to secure minimal dissemination of resistant helminths.(d) The study provided, for the first time, evidence regarding the association of industrial sites with a higher incidence rate of respiratory problems of lambs and kids. The findings may reflect that air pollution in farm environment can be a factor to take into account in health management. Possibly, data from farms can be employed to indicate potential risk from air pollution for humans; further work is necessary to draw relevant conclusions.(e) The study confirmed that high temperature can have an adverse effect on reproductive performance of sheep and goats. Notably, no adverse effects were seen in farms, in which hormonal administration was practiced as part of reproductive control procedures.(f) The study presented, for the first time, that climatic factors may affect bacterial numbers and their dissemination on teatcups from milking parlours. A model was developed for the study of staphylococcal kinetics on teatcups. Isolates of S. epidermidis have increased growth and dissemination on teatcups for sheep (silicone) than on teatcups for cattle (rubber). Environmental factors clearly affect the bacterial populations on teatcups in milking parlours, hence increasing the risk for mammary infections.(g) The study presented, for the first time, that climatic factors may overshadow factors traditionally considered for formulation of vaccination programs in sheep and goat farms. The findings are particularly useful within the context of precision medicine. The associations recorded may not be always the direct effect of climatic factors on disease outbreaks and may not reflect an association of the need to protect against infection outbreaks as climatic conditions change; they may also be the results of perceptions of farmers, as well as of consequences of an indirect influence of climatic factors on disease patterns.(h) Animal species other than sheep or goats are part of the farm ecosystem.(i) There are interactions between management of farm dogs and that of livestock on the farms; some farmers neglected both livestock and farm dogs, which may have influenced the health of all animals on their farms. The interactions are reflected in the omission of anthelmintic treatments of farm dogs and livestock. Hookworms and T. canis were found to be the most frequently detected helminths in farm dogs. An association was seen between presence of canid predators near the farms and parasitic infections in farm dogs.(j) The study provided a mapping of interactions of wildlife with small ruminants; presence of wildlife near small ruminant farms can adversely affect sheep and goat health. This also has consequences in the health management of farms, e.g., in anthelmintic treatment or in vaccination patterns. Increasing biosecurity measures at livestock farms and presence of livestock guard dogs could minimise predation risk, whilst also improving livestock welfare by reducing predator-associated stress.(k) Climate-related variables are associated with the quality of bulk-tank milk produced in the farms. These findings can reflect effects of climate-related variables on the animals producing milk or or direct effects on pathogens or effects to situations entirely outside the farm ecosystem. In any case, management-related variables or other elements of the ecosystem (e.g., wildlife animals or other domestic animals on the farms) can also be of importance.(l) Prediction models were developed, for the first time, for the incidence rate of abortion and lameness and for the presence of C. burnetii in the bulk-tank milk, by taking into account climatic conditions. Machine learning algorithms can be usefully employed in predicting infections in dairy small ruminant farms. Machine learning can take into account routinely available data, in order to provide predictions regarding financially important problems of sheep and goa. The findings will facilitate setting up appropriate measures and making interventions in small ruminant farms. Future research could explore further areas, for example, fine-tuning model parameters and incorporating additional data sources to improve the output of the model. In all, machine learning can advance sheep and goat farming by addressing challenges, improving decision-making processes, and enhancing veterinary clinical work and professional outputs. The entirety of the data will help to understand the ecosystem of sheep / goat farms in Greece, to provide information about the situation in the farms, to provide efficiency for time, to support development of relevant technologies and to optimise health management practices. Computational prediction models for the health management of farms can be developed routinely in the future. These can show excellent performance for various parameters related to health and milk production parametres. These models can be used as adjunct tools to support the decisions of clinicians in setting up management plans to improve animal health and welfare, as well as the quality of milk produced in small ruminant dairy farms. Along with further work from our group and with work of other researchers in Greece and internationally, ultimately it may thus be created a vision for the future of the small ruminant industry in Greece.
Degree
thesis:*- Grantor dc:publisher
- University of Thessaly (UTH)
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Katsarou, Eleni
- Κατσαρού, Ελένη
Subjects
dc:subject × 6Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/58626