Community surveys through space and time: testing the space-time interaction in the absence of replication
报告题目：Community surveys through space and time: testing the space-time interaction in the absence of replication
系列：“生态与环境讲坛”系列第 138 讲
摘要：Ecosystem monitoring requires repeated surveys through time (T). The same spatial (S) sampling design must be used in each survey. The surveys are usually done without small-scale replication through space; this allows the study to maximize spatial coverage. Classical statistics says that in such cases, one cannot test the space-time (S-T) interaction. But that interaction is very interesting, in particular in the framework of climate change: if the spatial structure of the data changes from year to year, it means that the climate has modified the ecosystem in important ways. The signature would be a S-T interaction, which cannot be tested unless replicate observations are available. We found a way to test the S-T interaction in the absence of replication. The method works with multivariate response data (e.g. species assemblages) as well as with a single response variable. The seminar will present the theory underlying this method. Then, results of numerical simulations will be mentioned, showing that the statistical test has correct rates of type I error. Two ecological applications of the method will be discussed.