Site-specific investigation and spatial modeling of canopy drip effect on element concentrations in moss

Abstract

In this study, the canopy drip effect on the exposure of forests to atmospheric deposition of potentially toxic metals and nitrogen (N) and element accumulation was investigated. Thereby, the respective element concentrations of metals and N in moss specimens were investigated by example of North-Western Germany. To this end, on the one hand, the concentrations of Al, As, Cd, Cr, Cu, Fe, Hg, Pb, Ni, Sb, V, Zn, and N in mosses sampled under, outside, and at the edge of forest canopies were examined for statistical significant differences. On the other hand, vegetation structures parameterizing the canopy drip effect were quantified by use of information collected, in addition to the element data, at each moss sampling site. The statistical relations between ratios of vegetation parameters and ratios of element concentrations were modeled by regression analysis, and the respective element concentration in moss was geostatistically estimated and mapped for unsampled locations throughout Germany. This article tackles regression models with R² > 0.5 (Cu, Hg, Pb, Sb, and N) to adapt the element concentrations measured at the 400 sites of the European Moss Survey (EMS) to three different features of hypothetical vegetation structures. To this end, the continuum of vegetation structures were represented as follows: open land (meadows) described by a leaf area index (LAI) value of 2.96 and under canopy sites in coniferous forests represented by a LAI value of 11. The arithmetic mean of LAI values at 400 EMS sites throughout Germany amounts to 5.1. The element concentrations for these target LAIs representing three site categories were calculated and mapped. Then, these LAI-dependent element concentration maps were compared with the maps depicting the spatial patterns of "pure" element concentrations. Spatial differences were evaluated and supposed to be of great value for the validation of atmospheric deposition modeling.