Growth and survival as a function of neighborhood density and
dbh
A linear, multiple regression was used to relate growth to six
variables: four measures of local neighborhood density, and elevation.
Trees were divided into four size categories, 2-5, 5-10, 10-20, and 20+
cm dbh (each category open on the right, so ≥ 5 and < 10,
etc.). This allowed each size category to have different responses. Stem
diameter was included as a predictor within each category because growth
changes considerably with size, especially in small trees. Growth was
calculated in five census intervals, each lasting five years, and the
census was included in the model as a random effect, accounting for the
repeated growth measurements of the same individual trees. The models
were run separately in the two species, Q. macrocarpa andQ. ellipsoidalis , meaning there were eight models all told, four
dbh categories in two species. The survival model was parallel in all
aspects, but with logistic regression replacing linear regression.