Applications of spatially explicit analyses in tanoak/redwood ecosystems

This dissertation presents three applications of spatially explicit analyses in tanoak (Lithocarpus densiflorus)/redwood (Sequoia sempervirens) stands. The first application utilizes ESRI GIS software to quantify the spread of Phytophthora ramorum symptoms on four plots in Marin County, California. The analysis results for the four plots indicate an average median spread of 2.1 meters and a range of increased density of symptomatic stems between 32 to 100 stems per hectare. Using Ripley's L(t) function and directional variograms I identify the scale of spread and find a pattern inconsistent with the pattern we would expect with wind dispersal of airborne inoculum. The second application examines the presence of disease symptoms using the same data set. I test the significance of traditional stand-level variables such as dbh, height, crown class, foliar condition; variables indicating the presence of insect, fungal infection or disease symptoms and spatially explicit indices derived from location, distance and density attribute data in a binary response multiple regression model. Simplified down from a maximal model of all variables following Akaike's Information Criterion the final model presented in this analysis includes only variables significant at p = 0.01. Significant spatial variables include the distance to the nearest infected tanoak stem and the percent basal area of tanoak stems within a 10 meter radius. The third application presents a spatially explicit analysis of redwood growth in a young even-aged plantation. Working with data from 15 plots along the Eel River in Humboldt County, Ca. I evaluate a number of individual stem and localize stand-level variables for significance in multiple regression analysis. In addition to looking at traditional variables such as dbh, height, and crown class, I examine sapwood cross-sectional area and crown-length. Using ESRI GIS software I derive spatial indices for each stem from location and individual attribute data and built a maximal model of variables simplifying this down following Akaike's Information Criterion. The final model includes only variables significant at p < 0.0001. The significant spatially explicit indices included in this final model account for the effect of above ground competition and microsite conditions on redwood growth.