P-224 - Predicting the potential risk of red ring nematode establishment using palm distribution in the United States

Abstract Text: The red ring nematode (Bursaphelenchus cocophilus), is a plant pathogen causing serious damage to palms and is present in Mexico, Central America, South America, and the Caribbean. While the nematode is absent in the contiguous U.S., its vector, the South American palm weevil (Rhynchophorus palmarum), has been detected in California and Texas. Limited data exist to determine the establishment of RRN within U.S. Thus, its potential distribution was modeled using palm trees (family Arecaceae) as a proxy. Using palm records from the Global Biodiversity Information Facility (GBIF), and bioclimatic variables and elevation data as predictors, nine correlative species distribution models (SDMs), logistic regression, general additive model, naïve bayes, classification and regression tree, random forest, k-nearest neighbor, support vector machine, artificial neural network, and maximum entropy, were trained. Optimal cutoffs were determined by maximizing Youden’s J, slightly improving some of the models’ test performance when compared to a default threshold of 0.5. Overall, the random forest model performed best with an overall accuracy = 0.967, Kappa = 0.934, Youden’s J = 0.934, and AUC = 0.993, using a threshold of 0.5. The mean temperature of the coldest quarter and mean precipitation of the warmest quarter were identified as the most important bioclimatic variables. Model predictions across the contiguous U.S. showed that much of the western and southern parts of the country, where palms are native and cultivated, could be at risk of red ring nematode establishment.