The relationship between average Ellenberg's indicator values per vegetation releve and environmental values measured in the field shows a large variability. This variability might have a strong impact on critical loads determined by dynamic models (i.e. by modelling the effect of acid deposition on the vegetation composition). To derive critical loads, acidity ranges per vegetation type are determined using Ellenberg indicator values, and subsequently translated into soil pH values. Acidity ranges are here defined as the range in which 80% of the characteristic species of a vegetation type can potentially occur. In this paper we explore the uncertainty in the modelled pH ranges and discuss the possible implications for critical loads. The effect of an overall translation instead of transfer functions per class on the estimated critical soil pH, and therefore on the critical loads, was calculated for a set of vegetation types in The Netherlands. The use of these different transfer functions resulted in critical soil pH values that differed by 0.2 pH units on average. For 12 out of the 17-investigated vegetation types this effect is of the same magnitude as the modelled effect of a 50% reduction in acidifying deposition. We also calculated the intrinsic uncertainty in the critical load estimates (i.e. due to the uncertainty in the regression parameters), which turned out to be even larger than the uncertainty due to the use of different transfer functions. This type of uncertainty was smaller when transfer functions per vegetation class were used (95% confidence interval c 1.4 vs c 1.7 pH units). Although the modelled critical loads have a large amount of uncertainty, they are not much different from experimentally derived critical loads. We conclude that modelled critical loads can be used when no empirical data are present, although they have to be,regarded as less certain. The uncertainty can be reduced by incorporating transfer functions per vegetation class.
- terrestrial ecosystems