A STATE-SPACE MODEL FOR UNIVARIATE ORDINAL-VALUED TIME SERIES

A state-space model was developed to analyze a univariate time series of ordinal-valued flower phenology data. Flower abundance, recorded as either “none,”“some,” or “much,” was observed each month on trees in a lowland Costa Rican rain forest to investigate flowering patterns of different species. Data from a single Capparis pittieri and daily rain measurements are used to demonstrate the model. A method to obtain maximum likelihood estimates of the parameters and the use of predicted probability differences to assess goodness of fit are described. Opportunities for improving the model are also discussed.