An objective procedure to remove observer-bias from phytoplankton time-series

Microscopic analyses of phytoplankton samples from the North Sea were made successively by three different laboratories in 1990 to 1993, 1994 to 1999 and 2000 to 2005. Analysis of the data from monitoring station Nw2 by multivariate ordination revealed that the plankton communities significantly clustered in three groups according to analysing laboratory. Furthermore, in each group a change in community structure was caused by an annual increase in the number of species observed. To remove these observer effects an objective five-step procedure was devised and tested using multivariate statistics. The result is a strongly reduced but still taxonomically diverse dataset with an ordination that is unrelated to the three analysing laboratories. After validating the five-step procedure with two other stations, it is concluded that this procedure is widely applicable, opening the possibility of comparing different time-series from countries along the North Sea.