Forest ecosystems can store massive amounts of organic carbon (OC) deep in their soils. Soil OC (SOC) is composed of several complex organic compounds of which groupings of these organic compounds can provide an insight into evaluating the formation and fate of SOC. However, there is little information available regarding the chemical composition of deep forest SOC and how they compare to upper soil depths. The objective of this study was to explore the SOC chemistry changes with increasing soil depth on a fine spatial scale in a planted forest soil derived from young volcanic erupted airfall deposits. Soil core samples to 1 m depth were collected from Puruki Experimental Forest (New Zealand) and examined for incremental depth changes in SOC functional groups, or chemical shift regions, using solid-state 13C cross polarisation magic angle spinning nuclear magnetic resonance (13C CPMAS-NMR) spectroscopy. The results showed that soil depth was a driver of change in soil chemical shift regions. The presence of a coarse-textured soil horizon, lapilli layer, at depth was a driver of significant (p < 0.05) differences between SOC functional groups. In the upper soil profile (0–10 cm), the chemical shift distribution was dominated by O-alkyl C (39%) followed by alkyl C (32%), aromatic C (23%), and carboxyl C (6%). In the deep soil layers, we found alkyl C increased below the lapilli layer (varying depth > 50 cm) compared to above the lapilli layer. In the deep soil layers the alkyl C was the dominant functional group (55%), followed by O-alkyl C (28%). Furthermore, there was a decrease at depth in aromatic (12%) and carboxyl C (5%). The A/O-A ratio increased with depth from soil above the lapilli layer compared to soils below the lapilli layer indicating a greater degree of organic matter decomposition and a corresponding decrease in the labile OC fractions. Within the lapilli layer the SOC functional groups and A/O-A ratio were highly variable. These results highlight that changes in SOC chemical composition occur with increasing soil depth, a property of deep soil that is frequently understudied, and are highly influenced by the multi layered nature of soil derived from volcanic airfall deposits.

Efficient water management plans are required to ensure food security in areas prone to drought and soil degradation. Hydrological modeling can help to achieve this goal by evaluating water availability for different uses, including food production. Fully distributed hydrological models using digital maps of soil properties available for the soil profile are useful for mapping soil-water balance responses, namely water content and water stress. The accuracy of these model responses depends on the quality of the input data, namely, soil hydraulic properties (SHP). Nowadays, we have free access to worldwide digital 1D SHP grids created using global soil datasets, so detailed studies are now needed to evaluate their use in hydrological modeling. We compared the use of a global and a local SHP dataset to simulate soil-water balance at the regional scale in a small irrigation district located in southern Portugal. We explicitly characterized where differences in model responses occur and related those to the variability of soil physical and hydraulic properties, and land use. Our results showed that the choice of the SHP dataset matters. The statistical and spatial variability of simulated water content and water stress for different land uses depended on the dataset used for the numerical simulation. With the global dataset, water content was higher and water stress lower, and both had less statistical and spatial variability compared to the local dataset results. A practical consequence of these findings is misadjusted irrigation schedules, which can be detrimental to crop production and soil quality. A water management plan based on the global results will determine less water requirements, which may not be realistic for the Mediterranean conditions that characterize the study area.

The origin of the soils formed on carbonate rocks remains uncertain, as they are probably of polygenetic origin. Of particular interest are the elevated contents of some trace elements detected in these soils, as they can hardly be attributed to insoluble residues of carbonate rock. The aim of this study was to uncover the relationship between heavy minerals in bedrock and soil and to identify other sources that influence the mineral and chemical composition of soil. We investigated representative samples of soils and dolomite bedrock as well as sandstones and marlstones from the nearby flysch basin as an expected source of the aeolian contribution. XRD and SEM/EDS showed that mineral diversity is higher in soils compared to the dolomites. Heavy minerals found in dolomite insoluble residue include sphalerite, Ti-oxide (probably rutile), zircon, fluorite, pyrite, minerals of REE phosphates and apatite group, interpreted as terrigenous detrital material. The mineral composition of soil heavy fraction is only partly following insoluble residues of bedrock and indicates possible aeolian contribution. Comparison with nearby flysch sandstones and marlstones showed similarities in mineralogical diversity by the presence of chromite and Fe-Cr(Mn) oxides (probably carmichaelite). Other minerals present in soils and flysch layers, such as Ti-oxide (probably rutile), zircon, REE phosphates, and pyrite grains, however, show similarities in mode of occurrence and transport indications. Signs of aeolian transport on the grains detected in soils were further confirmed by SEM/EDS.

Digital soil mapping techniques represent a cost-effective method for obtaining detailed information regarding the spatial distribution of chemical elements in soils. Machine learning (ML) algorithms using random forest (RF) models have been developed for classification, pattern recognition and regression tasks, they are capable of modelling non-linear relationships using a range of datasets, identifying hierarchical relationships, and determining the importance of predictor variables. In this study, we describe a framework for spatial prediction based on RF modelling where inverse distance weighted (IDW) predictors are used in conjunction with ancillary environmental covariates. The model was applied to predict the total concentration (mg kg−1) and assess the prediction uncertainty of 56 elements, soil pH and organic matter content using 466 soil samples in western Kenya; the results of iodine (I), selenium (Se), zinc (Zn) and soil pH are highlighted in this work. These elements were selected due to contrasting biogeochemical cycles and widespread dietary deficiencies in sub-Saharan Africa, whilst soil pH is an important parameter controlling soil chemical reactions. Algorithm performance was evaluated determining the relative importance of each predictor variable and the model's response using partial dependence profiles. The accuracy and precision of each RF model were assessed by evaluating out-of-bag predicted values. The models R2 values range from 0.31 to 0.64 whilst CCC values range from 0.51 to 0.77. The IDW predictor variables had the greatest impact on assessing the distribution of soil properties in the study area, however, the inclusion of ancillary environmental data improved model performance for all soil properties. The results presented in this paper highlight the benefits of ML algorithms which can incorporate multiple layers of data for spatial prediction, uncertainty assessment and attributing variable importance. Additional research is now required to ensure health practitioners and the agri-community utilise the geochemical maps presented here for assessing the relationship between environmental geochemistry, endemic diseases and preventable micronutrient deficiency.