Vegetation reflectance spectroscopy for biomonitoring of heavy metal pollution in urban soils

Abstract

Heavy metals in urban soils may impose a threat to public health and may negatively affect urban tree viability. Vegetation spectroscopy techniques applied to bio-indicators bring new opportunities to characterize heavy metal contamination, without being constrained by laborious soil sampling and lab-based sample processing. Here we used Tilia tomentosa trees, sampled across three European cities, as bio-indicators i) to investigate the impacts of elevated concentrations of cadmium (Cd) and lead (Pb) on leaf mass per area (LMA), total chlorophyll content (Chl), chlorophyll a to b ratio (Chla:Chlb) and the maximal PSII photochemical efficiency (Fv/Fm); and ii) to evaluate the feasibility of detecting Cd and Pb contamination using leaf reflectance spectra. For the latter, we used a partial-least-squares discriminant analysis (PLS-DA) to train spectral-based models for the classification of Cd and/or Pb contamination. We show that elevated soil Pb concentrations induced a significant decrease in the LMA and Chla:Chlb, with no decrease in Chl. We did not observe pronounced reductions of Fv/Fm due to Cd and Pb contamination. Elevated Cd and Pb concentrations induced contrasting spectral changes in the red-edge (690–740 nm) region, which might be associated with the proportional changes in leaf pigments. PLS-DA models allowed for the classifications of Cd and Pb contamination, with a classification accuracy of 86% (Kappa = 0.48) and 83% (Kappa = 0.66), respectively. PLS-DA models also allowed for the detection of a collective elevation of soil Cd and Pb, with an accuracy of 66% (Kappa = 0.49). This study demonstrates the potential of using reflectance spectroscopy for biomonitoring of heavy metal contamination in urban soils.