Detecting livestock production zones.

Communities are sets of nodes that are related in an important way, most likely sharing common properties and/or playing similar roles within a network. Unraveling a network structure, and hence the trade preferences and pathways, could be useful to a researcher or a decision maker. We implemented a community detection algorithm to find livestock communities, which is consistent with the definition of a livestock production zone, assuming that a community is a group of farm premises in which an animal is more likely to stay during its lifetime than expected by chance. We applied this algorithm to the network of animal movements within the state of Mato Grosso for 2007. This database holds information concerning 87,899 premises and 521,431 movements throughout the year, totaling 15,844,779 animals moved. The community detection algorithm achieved a network partition that shows a clear geographical and commercial pattern, two crucial features for preventive veterinary medicine applications; this algorithm provides also a meaningful interpretation to trade networks where links emerge based on trader node choices.

Prevalence and risk-mapping of bovine brucellosis in Maranhão State, Brazil.

Between 2007 and 2009, a cross-sectional survey was carried out in Maranhão State, Brazil to estimate the seroprevalence of and risk factors for bovine brucellosis. In total, 749 herds and 6779 cows greater than two years of age were blood sampled. At the time of sampling a questionnaire to collect details on possible risk factors for bovine brucellosis was administered to the participating herd manager. A logistic regression model was developed to quantify the association between herd demographic and management characteristics and the herd-level brucellosis status. Spatial analyses were carried out to identify areas of the state where the presence of brucellosis was unaccounted-for by the explanatory variables in the logistic regression model. The estimated herd-level prevalence of brucellosis in Maranhão was 11.4% (95% CI 9.2-14) and the individual animal-level prevalence was 2.5% (95% CI 1.7-3.6). Herds with more than 54 cows older than two years of age, herds that used rented pasture to feed cattle, and the presence of wetlands on the home farm increased the risk of a herd being brucellosis positive. Infected farms were identified throughout the state, particularly in the central region and on the northwestern border. Spatial analyses of the Pearson residuals from the logistic regression model identified an area in the center of the state where brucellosis risk was not well explained by the predictors included in the final logistic regression model. Targeted investigations should be carried out in this area to determine more precisely the reasons for the unexplained disease excess. This process might uncover previously unrecognized risk factors for brucellosis in Maranhão.