Long-term high resolution vegetation data – analytical approaches

We present different approaches to analyse long-term, high resolution forest vegetation data from the ICP Integrated Monitoring network, to assess vegetation dynamics and possible causes. Our data consists of plant species cover from permanent monitoring plots of about 40 by 40 m, from 40 sites in 14 European countries. Each site has one or two monitoring plots. The vegetation is monitored in about 30 subplots of 50 x 50 cm, randomly located within a plot. For analyses of temporal trends, we removed plots with less than 4 inventories over time, resulting in data from only 8 plots in 3 countries. Multivariate analyses using distance moved in ordination space to track temporal trends in species composition at the plot level show that some plots are very stable while others show a high species turnover as responses to perturbations in the environment, mainly bark beetle infestations. We also conclude that some affected plots return to a pre-disturbance species assemblage, indicating a high resilience. Plant community stability was also assessed as the temporal mean divided by the temporal standard deviation, resulting in an index of temporal stability. Results show that plots with a high stability index also show consistent high stability in the multivariate analyses of temporal plot stability. Environmental changes reflected in the plant communities were assessed by Ellenberg indices for soil acidity, nutrients, and temperature, calculated as weighed means, followed by temporal trend analyses using Theil-Sens’s slope. Among the plots that have reported sufficiently long time series, we see a small but significant decrease in plant’s preference for soil acidity in Swedish plots. However, most plots included in these analyses did not show any temporal trends in tested Ellenberg indices. Changes in temporal frequency for individual species among subplots within plots was assessed by generalized linear models (GLMs), using a logit link function and a binomial error distribution. Only a small fraction of the species showed a significant change in fraction of subplots where they were present. Significant changes were mainly restricted to mosses. One conclusion from this study is that even with rigorous monitoring protocols and reporting templates followed by thorough data quality assessment before storage in a common database as for ICP IM, there are nevertheless data issues that need to be addressed before any analyses. This shows that the intense work with eLTER Standard Protocols and data handling is of utmost importance.