For the evaluation of the epidemiology of Theileria equi and Babesia caballi in a herd of 510 horses in SW Mongolia, several mathematical models of the transmission dynamics were constructed. Because the field data contain information on the presence of the parasite (determined by PCR) and the presence of antibodies (determined by IFAT), the models cater for maternal protection with antibodies, susceptible animals, infected animals and animals which have eliminated the parasite and also allow for age-dependent infection in susceptible animals. Maximum likelihood estimation procedures were used to estimate the model parameters and a Monte Carlo approach was applied to select the best fitting model. Overall, the results are in line with previous experimental work, and add evidence that the epidemiology of T. equi differs from that of Babesia spp. The presented modelling approach provides a useful tool for the investigation of some vector-borne diseases and the applied model selection procedure avoids asymptotical assumptions that may not be adequate for the analysis of epidemiological field data.Key words: Theileria equi, Babesia caballi, horses, tick-borne disease, epidemiology, Mongolia, mathematical modelling, Monte Carlo methods.
I N T R O D U C T I O NEquine piroplasmoses are caused by two intraerythrocytic protozoa, Theileria equi and Babesia caballi. Both are transmitted by ixodid ticks (Friedhoff, 1988). Clinical signs of infection may vary from asymptomatic to acute fever, anaemia and dyspnoea, and even death (reviewed in Schein, 1988). Chronically infected horses represent a reservoir infecting ticks, which subsequently transmit the parasites to other equids. Piroplasms can be detected directly by microscopical examination of Giemsa-stained thin blood smears or by polymerase chain reaction (PCR) (Bruning, 1996), and for indirect diagnosis, the immunofluorescence antibody test (IFAT) is the most widely used technique (Gummow et al. 1996 ;Avarzed et al. 1997 ;Heuchert et al. 1999).The epidemiology of equine piroplasmoses has been investigated in various studies (Mahoney, 1969 ;Mahoney and Ross, 1972 ;Ross and Mahoney, 1974 ;Smith, 1983 ;Dallwitz et al. 1987 ;Medley et al. 1993). In 2004, a study was conducted in a domestic horse population in south-western Mongolia (Rü egg et al. 2007). PCR results indicated a T. equi prevalence of 66 . 5 % (95 % CI : 62 . 2-70 . 7) and the IFA test demonstrated that 78 . 8 % (95 % CI : 74 . 9-82 . 3) of animals had seroconverted to T. equi. The corresponding values for B. caballi were 19 . 1 % (95 % CI : 15 . 7-22 . 8) and 65 . 7 % (95 % CI : 61 . 4-69. 9) respectively. To investigate the impact of age, herd affiliation, sex, date of sample collection and tick abundance on the PCR and IFAT prevalences, a generalized linear model (GLM) and a generalized additive model (GAM) were used. In both models, sex and age were the only two significant explanatory variables (Rü egg et al. 2007). Despite being useful for detection of risk factors and testing hypotheses about relations...