Predicting cattle movement networks: Mycobacterium bovis spatial-genotyping versus gravity modelling

Understanding cattle movement patterns allow us to predict hotspots and develop targeted control strategies to save scarce resources. This study therefore compares the utility of molecular and census data in predicting cattle movement networks in Cameroon. Cattle movement network predictions were made using two approaches: a molecular-based Network (MN) by transforming phylogenetic distances of Mycobacterium bovis (Spoligotype &MIRU-VNTR) genotypes recovered from an abattoir-based study across four regions of Cameroon conducted in 2013; and a gravity model network (GN) based on the ³Cattle/Human² ratio retrieved from 2005-2007 census data and the Euclidean distances computed from GIS data from Cameroon was generated. The models were validated by comparing the predictions to an empirical network (EN) generated using data on cattle sources at livestock markets located in the four regions collected in 2014. Statistical, spatial and network measures of centrality comparisons were conducted using the ggmap and igraph packages in R version 3.0. The three networks showed similar spatial patterns of cattle movements. We however observe more consistency of animal flow across MN and EN from the North Region towards the Adamawa and North-West Regions. The measures of centrality in the EN indicate that Ngaoundere, Nyambaka, Nankoue and Ngaoundal are prominent actors in the cattle movement network, three and two of which were also predicted by the MN and the GN respectively. There is better statistical and spatial agreement between EM & MN than GN, which suggests that molecular genotypes are better predictors of cattle movement networks in Cameroon.