This item has been officially peer reviewed. Print this Encyclopedia Page Print This Section in a New Window This item is currently being edited or your authorship application is still pending. View published version of content View references for this item

Results

Authored By: M. C. Downing, S. T. Jung, V. L. Thomas, M. Blaschke, M. F. Tuffly, R. M. Reich

For the original Bavarian model: of the 19,620 forested square kilometers assessed in Bavaria, approximately 14,015 square kilometers (71.43 percent) were modeled to have a high potential for P. alni root and collar rot, and 5,604 square kilometers (28.56 percent were modeled to have a high potential to remain healthy (Figure 1).

Seven terminal end nodes were used and accounted for 78.34 percent of the variability. The tenfold cross validation gave a higher accuracy for predicting the 307 P. alni infested sites at 86 percent, and showed 63 percent accuracy in predicting the 127 healthy sites. The independent variables important in predicting the presence or absence of P. alni were minimum silt fraction values less than 20 percent (range = 0 to 80 percent), mean sand fraction values less than 5 percent (range = 0 – 93 percent), slope less than 2.97 degrees (range = 0 to 30.74 degrees), aspects that were Southeast, South, Southwest, and West, and the landform index less than 6.6 (range = - 15.20 to + 21.60; <0 = concave; 0 = totally flat; >0 = convex) (Figure 2).

For the second Bavarian model, with only the three independent variable datasets that are available globally being used, fourteen terminal end nodes accounted for 77.19 percent of the variability. The tenfold cross validation still showed a higher accuracy for predicting the 307 infested sites at 92.51 percent, but only 40.16 percent accuracy in predicting the 127 healthy sites. As expected, all three of the independent variables were used to predict the presence or absence of P. alni, but, in this model, slopes less than 19.57 degrees had a higher probability for infestation, and the East aspect was selected in addition to the four aspects that had originally been selected. Also, landform indexes less than 6.2 had a higher probability of infestation.

The final global susceptibility surface had 27,835,766 square kilometers of suitable area where alder and P. alni could survive (Figure 3). Of that area 1,482,487 square kilometers (5.33 percent) were highly susceptible to P. alni; 3,930,660 square kilometers (14.12 percent) had a medium susceptibility; 5,721,467 square kilometers (20.55 percent) had a low susceptibility; and 16,701,152 square kilometers (60.00 percent) had little or no susceptibility (Table: Phytophthora alni susceptibility).

Encyclopedia ID: p3331



Home » Environmental Threats » Case Studies » Case Study: Estimating the Susceptibility to Phytophthora alni Globally » Results


 
Skip to content. Skip to navigation
Text Size: Large | Normal | Small