Tidal marshes are coastal ecosystems that store large amounts of sedimentary organic carbon (OC). Reducing the current uncertainty on the amount of OC stored these sediments requires the analysis of a large number of sediment samples. Soil sensing techniques, using mid infrared (MIR) spectroscopy combined with partial least squares regression (PLSR), have been proven to be a valid alternative for standard OC measurements in a wide range of terrestrial ecosystems. However, the application of these techniques to tidal marsh sediments has been very limited and the error associated with calculated sedimentary OC stocks using MIR spectroscopy/PLSR data has up till now not been evaluated. Therefore, we assessed the potential of MIR spectroscopy/PLSR to predict the OC concentration and OC stock of tidal marsh sediments in a temperate estuary (Scheldt estuary, Belgium and The Netherlands). Based on a cross-validation procedure, the results show that MIR spectroscopy/PLSR predicts the OC concentration of tidal marsh sediments along the entire estuary with a high accuracy (R2 = 0.94, RMSE = 0.56% OC). Similar accuracies were obtained for the prediction of sedimentary OC concentrations from independent tidal marshes. Organic carbon concentrations from tidal marshes in the salt portion of the estuary were, however, predicted more accurately using a PLSR model trained exclusively with sediment samples from this salinity zone. Combining depth profiles of predicted OC concentrations with measured bulk densities resulted in predictions of total OC stocks with a relative error < 4% for freshwater and brackish marshes, while errors for salt marshes were up to 30%. As MIR spectroscopy/PLSR is not able to accurately predict bulk density or OC density, standard field measurements of bulk density remain necessary to reliably predict sedimentary OC stocks in tidal marshes.
Alle informatie in het Integrated Marine Information System (IMIS) valt onder het VLIZ Privacy beleid