Using classification tree analysis to predict oak wilt distribution in Minnesota and Texas

Abstract

We developed a methodology and compared results for predicting the potential distribution of Ceratocystis fagacearum (causal agent of oak wilt), in both Anoka County, MN, and Fort Hood, TX. The Potential Distribution of Oak Wilt (PDOW) utilizes a binary classification tree statistical technique that incorporates: geographical information systems (GIS); field sample data; commonly available, inexpensive, coarse-resolution auxiliary data; and satellite imagery from both Landsat Thematic Mapper (TM) and SPOT to predict the spatial distribution of oak wilt. Two types of model evaluations were conducted -- a ten-fold cross validation and an assessment using additional oak wilt data that had been verified in the field. These evaluations indicated that at the landscape scale PDOW correctly models the presence of oak wilt, and accurately predicts oak wilt distribution in Anoka County, MN and Fort Hood, TX.

Citation

Downing, Marla c.; Thomas, Vernon L.; Juzwik, Jennifer; Appel, David N.; Reich, Robin M.; Camilli, Kim. 2008. Using classification tree analysis to predict oak wilt distribution in Minnesota and Texas. In: Billings, Ronald F.; Appel, David N., eds. The proceedings of the 2nd national oak wilt symposium; 2007 June 4-7; Austin, TX. National Oak Wilt Symposium: 67-83.