On reproductive and respiratory syndrome virus dissemination across pig production systems in the United States

Manuel Jara; David A. Rasmussen; Cesar A. Corzo; Gustavo Machado

Bayes factors (BF) (Dellicour et al., 2016). Based on the extracted spatio-temporal information contained 166 in each phylogenetic tree described above, we calculated the distances based on predictor variables 167 associated with every branch of each tree, which were used to determine the correlation between the 168 duration of each branch in the phylogeny and its related predictor variable distance, including the annual 169 temperature, annual precipitation, elevation, runoff (an index that quantity of water discharged in surface 170 streams), soil humidity (commonly used to quantify topographic control on hydrological processes) 171 (Sørensen et al., 2005) , enhanced vegetation index (EVI), which can be used to quantify vegetation 172 greenness, as well as the pig density, and distance to the main roads ( Fig. 1 and Table S3 for more 173 details). Briefly, the distance between each farm (geographic coordinates) to the nearest main road was 174 calculated based on the Euclidean distance. We calculated the statistic Q, which represents how much 175 variation in lineage movement is explained with spatial heterogeneity when each predictor variable is 176 considered (Dellicour, Rose and Pybus, 2016). Here, Q= R 2 env -R 2 null where R 2 env is the coefficient of 177 determination obtained from the regression between branch duration against predictor (factors) distances, 178 while R 2 null represents the coefficient of determination for the regression between branch duration against 179 predictor distances, which were replaced by a null raster. Thus, the factors that were considered for 180 further Bayes factor (BF) analysis, were the ones that obtained positive Q values in at least 90% of the 181 tested trees (Jacquot et al., 2017). We approximated a BF value for all predictors tested that passed the 182 above-mentioned Q statistic. Predictive factors were treated as a conductance (variables that promote the 183 spread of the disease) or a resistance factor (variables that impede its spread). For the interpretation of BF, 184 values between 3-20 were considered "positively" supported, values between 20-150 were considered 185 "strong", and values >150 were considered " overwhelming support" (Kass and Raftery, 1995 supplementary Table S1 ). 193