Dear list, I have a time series of frequency data (number of dead trees in each year), including zero frequencies. With a continuous predictor variable (climate index), I want to assess the effect on the number of dead trees. Is there a log-linear regression model that includes a serial correlation structure to model temporal dependence? I think of an autoregressive model, such as in the functions lme or gls in library(nlme). Any help is appreciated! Christof