Timber isoscapes. A case study in a mountain area in the Italian Alps

Abstract

Background: Local timber is still one of the main sources of work and income for mountain communities. However, illegal logging is a major cause of deforestation in many countries and has significant impacts on local communities and biodiversity. Techniques for tracing timber would provide a useful tool to protect local timber industries and contribute to the fight against illegal logging. Although considerable progress has been made in food traceability, timber provenance is still a somewhat neglected research area. Stable isotope ratios in plants are known to reflect geographical variations. This study reports accurate spatial distribution of δ18O and δ2H in timber from north-eastern Italy (Trentino) in order to trace geographical origin.

Methodology and principal findings: We tested the accuracy of four kriging methods using an annual resolution of δ18O and δ2H measured in Picea abies. Pearson's correlation coefficients revealed altitude to be the most appropriate covariate for the cokriging model, which has ultimately proved to be the best method due to its low estimation error.

Conclusions: We present regional maps of interpolated δ18O and δ2H in Picea abies wood together with the 95% confidence intervals. The strong spatial structure of the data demonstrates the potential of multivariate spatial interpolation, even in a highly heterogeneous area such as the Alps. We believe that this geospatial approach can be successfully applied on a wider scale in order to combat illegal logging.

Fig 1. Location of the 151 sampling sites in Trentino Province.

Altitude is represented by the white-green colour scale.

Fig 2. Box plots of δ¹⁸O and δ²H and the 4 candidate covariates.

Fig 3. Variograms of δ²H and δ¹⁸O for 2013 and 2014.

The A column shows those computed without trends, B indicates variograms detrended by a linear model, and C shows those detrended by a quadratic model. The lags have been binned over all directions and incremented in steps of 2,500 m. Solid lines show the models fitted to the variograms with the parameters reported in each panel. The fitted models were chosen on the basis of the smallest RSS, as reported in the text.

Fig 4. Anisotropy maps of δ¹⁸O and δ²H plotted for 2013 and 2014.

Fig 5. Variograms and cross-variograms of δ²H and δ¹⁸O and elevations.

The lags have been binned over all directions and incremented in steps of 2,500 m. Solid lines show the models fitted to the variograms with the parameters reported in each panel.

Fig 6. Scatter diagrams of δ²H and δ¹⁸O measured in 2013 and 2014 plotted against the values predicted by cokriging after cross-validation.

The solid line represents the regression. R² is reported for each plot.

Fig 7. δ²H isoscapes and relative 95% confidence intervals for Norway spruce timber for 2013 and 2014.

Fig 8. δ¹⁸O isoscapes and relative 95% confidence intervals for Norway spruce timber for 2013 and 2014.