The response variables: Ponatinib species richness of aspen-dependent lichens, cyanolichens, spore-dispersed lichens, lichens sensitive to light and lichens adapted to open environments were also analyzed with the same model. Four environmental variables were used as covariates: stand area (ha), aspen diameter (cm), latitude and longitude. Interactions between each covariate and age-class were included in the full model. Subtracting the mean and dividing with the standard deviation standardized all variables. To get an estimate of how well the models

explained the data, we used an information-theoretic approach based on likelihood measures, in our case the Akaike information criterion (AIC; Akaike, 1974), and to correct for small sample sizes we used AICc. Several models were similar regarding their AICc value, making it difficult to select the best model. Consequently, to handle the problem with model selection uncertainty, we used model averaging (Burnham and Anderson, 2002) where all models within ΔAICc ⩽ 4 were weighted. This Akaike weight can be interpreted as the probability that a model is the best among the candidate models. Each variable included in at least one of the models got an average parameter estimate, based only on the models within ΔAICc ⩽ 4 where the parameter was present, with a confidence interval and

a relative importance. The relative variable importance is the probability that a variable will appear in the best model, and is based buy NVP-BEZ235 on the models’ Akakie weights (Whittingham et al., 2006). The analyses were conducted in the statistical software R v.2.12.1 (R Development Core Team, 2010) using the lme4 package (Bates et al., 2011) for the full model and the MuMIn package (Barton, 2012) for model averaging. We used indicator species analysis (ISA) to identify characteristic species of the age classes. ISA is based on a method by Dufrêne and Legendre (1997) that combine information on how often a species is present in a group and how unique it is to that group. A perfect 2-hydroxyphytanoyl-CoA lyase indicator is always and only present in one group. Statistical

significance is tested through Monte Carlo randomizations (McCune et al., 2002). PC-ORD version 5.31 (McCune and Mefford, 2006) was used for the analysis. Rarefaction curves based on trees as samples (Gotelli and Colwell, 2001), and estimators of total species richness (Magurran, 2004) were used to assess the size of the regional species pool. We used three non-parametric estimators since they are less sensitive to the underlying shape of the species-abundance distribution than parametric estimators (Magurran, 2004), (1) Chao 2 (Chao, 1984). (2) Jackknife 2 (Smith and van Belle, 1984), (3) Bootstrapping (Smith and van Belle, 1984). Calculations were made in EstimateS (Colwell, 2005). We found 195 species in total, 131 on trees exposed for 0–4 years and 182 on trees exposed for 10–16 years.