LINEAR MODELING OF IRREGULARLY SPACED, UNBALANCED, LONGITUDINAL DATA FROM PERMANENT-PLOT MEASUREMENTS

A linear mixed-effects model that accounts for the covariances among r epeated measurements and for random plot effects is developed. A conti nuous-time autocorrelation error structure permits the model to be app lied to irregularly spaced, unbalanced data. A strategy for fitting th is model and diagnostic tools for assessing its goodness of fit are pr esented and demonstrated. The model is fitted to two permanent-plot da ta bases and shows marked improvement compared with models that do not account for the error structure.