Modeling the secondary spread of viral hemorrhagic septicemia virus (VHSV) by commercial shipping in the Laurentian Great Lakes

Researchers have only begun to study the role of shipping in the spread of invasive species in the Laurentian Great Lakes despite a well-documented history of introductions in these lakes due to ballast water release. Here, we determine whether ballast water discharge was a likely vector of spread of the fish disease, viral hemorrhagic septicemia virus genotype IVb (VHSV-IVb), throughout the Great Lakes and St. Lawrence Seaway. Three models were developed to assess whether the spread of VHSV was due to (1) chance (random model), or (2) ballast water discharge (location model), and whether (3) increased propagule pressure, as measured by the number of visitations by ships carrying ballast water from VHSV infected areas, increased the likelihood of a discharge location becoming infected with VHSV (propagule pressure model). The third model was also used to assess the probable point of initial introduction of VHSV. Presence and absence accuracies and weighted Cohen’s kappa were calculated to determine which models best predicted observed presences and absences of VHSV. Location models explain the patterns of VHSV detections better than random models, and inclusion of “propagule pressure” often improved model fit; however, the relationship is weak likely because of a long lag time between introduction and detection, a high rate of false negatives in reporting, and the possible contribution of other vectors of spread. Montreal was also identified as the more likely introduction site of VHSV, rather than Lake St. Clair, the site where the virus was first detected.