Pedological features affecting carbon accumulation in submerged-soils sequence assessed with Vis-NIR spectroscopy in a coastal wetland in Albania

• Carbon accumulation in submerged soils is affected by pedological features. • Hydromorphic conditions influence the C accumulation and plant communities variation. • C fractions were analyzed using dry combustion and Vis-NIR spectroscopy. • Vis-NIR spectroscopy linked pedological features and C accumulation. The sequestration of SOC in submerged soils depends on the anaerobic conditions and vegetation types. In this context, soil spectroscopy is an accurate and reproducible method to estimate SOC in the visible and near infrared region. The aim of this research was to quantify the SOC fraction in wetland soils using soil spectroscopy and, subsequently, to try and establish a relation of different carbon fractions with soil pedological features. A transect of 17 pedons was sampled to analyze pedological features that affect carbon accumulation and decarbonation processes in the Kune Vaini lagoon system in Albania. The organic and inorganic C pools were analyzed using dry combustion, with a CNS analyser to determine total organic carbon (TOC), and with a furnace to determine loss on ignition (LOI) at 350 and 450 °C. Vis-NIR spectra were analysed using two regression models (PLSR and SVMR) that were calibrated and validated to obtain a valuable prediction for soil C fractions. Results indicated that the difference of soil organic matter (SOM), estimated with CNS and LOI, can be considered to represent the SOM that is strongly resistant to mineralisation. The SOM from LOI450 was considered to have more resistant forms of organic C, in spite of the more labile and degradable ones (TOC, LOI350). Models derived from PLSR had slightly higher RPD values compared to the results obtained from the SVM regression modelling for all C fractions and for carbonate, EC, BD, and pH. This study showed that the use of visible near infrared spectroscopy coupled with the PLSR model allowed for the accurate prediction of LOI, TOC, CaCO 3 , BD, EC, and pH of submerged soils. The PLSR model can be applied in estimating different carbon contents under submerged conditions to get a deeper understanding of carbon accumulation in wetland soil ecosystems.