Fig. 18.17 Boxplot of log-transformed DIAT1 conditional on laboratory (represented by A, B, C, and D) per area factor Lab. As well as looking further into this particular DIAT1-group either on the level of species or per station, the Lab-pattern appeared not to be of a structural kind. Summarising, we cannot say whether any changes over time in abundances are due to a laboratory effect or whether it represents a real change. We therefore concluded that the laboratory effect is small compared to observed changes and ignored it.

The modelling approach followed similar lines used for the environmental variables. Note that most long-term trends in Fig. 18.18 seem to increase up to about 2001. Seasonal patterns are rather different per area (Fig. 18.19). Some areas show a clear diatom blooming in early spring followed by a smaller bloom in late autumn. The spatial pattern is given in Fig. 18.20.

Fig. 18.18 Estimated trends for log-transformed diatoms (DIAT1). The model did not contain a laboratory effect

Was this article helpful?

0 0

Post a comment