Application of spatial modeling for early detection of Sudden Oak Death in forest landscapes

Identifying the establishment of invasive organisms and monitoring their spread is essential for management and conservation of threatened habitats. Application of a recently developed spatial model was studied for its usefulness in guiding early detection surveys and evaluating broad-scale patterns of invasion by the exotic pathogen Phytophthora ramorum, causal agent of Sudden Oak Death. Within forests predicted as high risk for establishment and spread of P. ramorum, 890 site locations were randomly targeted across California for early detection over a 3-year period. Leaf samples were collected from hosts with P. ramorum symptoms along two transects per site. The pathogen was detected at 78 of the 890 locations assessed. Logistic regression indicates that P. ramorum is more likely to occur at a randomly located field site as proximity to known infections increases. The probability of pathogen presence did not differ among the three sampling years. The distance-decay logistic regression equation applied to a GIS predicts the pathogen to occur across 10.5% of California's 9,228 km super(2) of threatened forest, but with less than 0.1% of this area occurring beyond 20 km of a previously known infection. The exponential decay pattern indicates that long-distance dispersal of P. ramorum has occurred, but the majority of spread has occurred locally within and between forest stands during the 3 year sampling period. As disease continues to develop, identifying isolated infections associated with long distance dispersal will be key to managing large scale spread. The combination of random sampling and risk-based site selection appears to be an effective approach for early detection monitoring of disease distribution and outbreaks. This study is one of a few examples using a spatially-explicit model of invasive species spread for guiding management activities and conservation of native species.