Use of geographical information system and ecological niche modelling for predicting potential space distribution of subclinical mastitis in ewes

•Ecological Niche Modelling was employed to predict high risk areas of mastitis in sheep.•Data from an extensive countrywide field investigation in Greece were taken into account.•Differences were evident according to management system applied in farms.•Differences were evident for subclinical mastitis caused by any organism and subclinical mastitis caused by slime-producing staphylococci. Objective was to recognise areas potentially of high risk for increased frequency of subclinical mastitis in ewes. Milk samples were collected, for bacteriological and cytological examination, from 2198 clinically healthy ewes in 111 farms in all administrative regions of Greece. Data on farms were located in the field using hand-held Global Positioning System Garmin units. Collected data were analysed by an Ecological Niche Model under the framework of a geographic information system. Two separate analyses were performed: one for subclinical mastitis independently of causal agent (prevalence in population sampled: 0.260) and one for subclinical mastitis caused specifically by slime-producing staphylococci (prevalence in population sampled: 0.153). A model was constructed in which sheep farms were divided into two clusters, according to prevalence of subclinical mastitis: farms in the upper three quartiles of prevalence were used as occurrence points for the Ecological niche modelling procedure (‘infected farms’); farms in the lower quartile of prevalence within each category were (pseudo)negative points. Significant differences in environmental parametres prevailing in locations of farms into the study, were identified for up to 13 parametres between locations of farms according to management system applied in farms. When farms in each management system were considered separately, differences became evident between farms in each management system, as well as between the two infections. The factor with the highest relative contribution in the analyses was the distance from other sheep farms; other factors also of importance in the predictive models were the altitude, the maximum temperature of warmest month and the total precipitation of driest month. Verification of the model revealed that ≥ 0.760 of infected farms’ were located in areas predicted as high risk for prevalence of subclinical mastitis or slime staphylococcal subclinical mastitis. The paper describes for the first time potential association of mastitis with environmental factors and presents predictive mode