Using heterogeneity in the population structure of U.S. swine farms to compare transmission models for porcine epidemic diarrhoea

Eamon B. O’Dea; Harry Snelson; Shweta Bansal

What features of the time series might have driven the results in our correlation analysis? The correlation between transport and cross correlations seemed to be driven in part by concentration of both high cross correlations and large flows in Midwestern states ( Supplementary Fig. S1) . The cross correlations of these states results from the presence of a small wave of positive accessions early in the outbreak and a much larger wave toward the end of our observations (Fig. 3b, left column) . Also, Kansas and Oklahoma share a distinctive period of high positive accessions in the middle of the time series and fairly large flows (Supplementary Fig. S9 and Fig. 3b) . Epidemiological reports suggest that windborne transmission was important for spread in Oklahoma, and thus the summer activity in Kansas and Oklahoma may also have been a consequence of short-distance, spatial spread. It is not known how the spread in these states occurred in spite of the high temperatures that were thought to have slowed PEDV transmission in other states during the summer. North Carolina is distinct from the Midwest in that no outbreaks occurred in the spring and many outbreaks occurred in October (Fig. 3b) . Epidemiological reports suggest that PEDV was introduced in the Midwest and eventually reached North Carolina via transport. The high number of sow farms in the state may have allowed for a largely self-sustained cluster of outbreaks following introduction. Although the matrices measuring spatial transmission may not have come out as significant in our analysis, consideration of the epidemiological explanations for the time series suggests that spatial transmission may to some extent explain the significance of the transport flow matrix.