Investigation of permanent plots is the traditional approach to detect changes in species performance and floristic composition. When the time reserved for investigations is limited and statistically independent replicate samples for normal time series analysis do not exist, ordination of multi-species series is often applied. The approach is further developed here with time series data from wetland communities over six consecutive years. Random fluctuation and linear trend are the two mechanisms which can explain the observed changes. Trend analysis of species scores allows to smooth the data and hence the resulting ordination pattern. The expected scores are a conservative measure for trend, taking into account all the recorded time states of the system.