Correction to: Recalibration and validation of the Swiss lichen bioindication methods for air quality assessment.

The correct fig. 2 is presented in this paper.

Recalibration and validation of the Swiss lichen bioindication methods for air quality assessment.

The aim of this study was to recalibrate the Swiss lichen bioindication methods, developed and calibrated with air pollution data 30 years ago. Since then, levels of air pollution have considerably decreased, and the mix of pollutants has changed due to successful emission control in Switzerland and neighboring countries. In particular, there has been a change from a sulfur- and acid-dominated to a more nitrogen-dominated pollution mix of NH3/NOx and ozone, resulting in increased pH levels. This allowed a recolonization and change in abundance and composition of the epiphytic lichen vegetation, indicating an improved air quality in Switzerland. The existing indices of atmospheric pollution or purity IAP18 and IAPBR developed 30 years ago showed good longitudinal correlations with air pollutant levels until the end of the last century, but a growing drift was observed in some regions over the last 15 years. This called for a method recalibration with more recent air pollution data. Data from a total of 7178 trees from 22 Swiss regions grouped into 1331 homogenous plots and covering the period 1994 to 2017 were averaged by year within plots. Three pollutant-specific lichen indices were newly established, one for primary pollutants (NO2, PM10, SO2), one for ozone (AOT40f), and one for ammonia (NH3). These pollutant-specific lichen indices were derived from linear regression models with lichen variables and a linear time trend variable as predictors, using time-dependent coefficients. Parameters were selected using the Lasso method. The primary pollutant lichen index showed a coefficient of determination R2 of 0.86 in the model with NO2, PM10, and SO2 as predictor variables, whereas corresponding models with other predictor variables (i.e., NH3, AOT40f, and meteorological variables) were of considerably lower fit. Regionalized lichen models for three larger Swiss regions revealed even better results, compared with the unified Swiss models. The best regionalized ozone and ammonia lichen indices reached an R2 of 0.88 and 0.71, respectively.