Roland Bol

Acid-extractable concentrations of Cr, Mn, Cu, Zn, As, Cd and Pb were measured by ICP-MS in 80 composite topsoil samples collected from five agricultural sites in the West Bank, Palestine, near wastewater sources or a landfill. Pollution indices calculated relative to continental-crust values differed among sites but are conditional on the partial extraction and selected background. Mean sample-level PLI classified ADJ (1.12), EQR (1.23) and DSN (1.33) as polluted. Cd had the highest I _geo at ADJ (3.1), and Fe-normalized EF classified Cd at ADJ as extremely enriched. An Al-normalized sensitivity analysis preserved the principal Cd, As, Zn and Cr patterns but changed four of 35 site–metal EF classes. Mean RI was highest at ADJ (424.5; considerable screening risk), followed by DSN (150.7; at the low–moderate boundary); Cd contributed 93% and 73% of RI at these sites. A validated 5 × 4 self-organizing map produced four stable compositional clusters. Exploratory three-factor PMF identified Cr-rich soil-matrix, Cd–Zn-enriched mixed and Mn-dominant signatures, although residual over-dispersion, rotational ambiguity, weak reconstruction of Cu, Zn and Pb, and site dependence precluded quantitative source attribution. These screening results identify Cd, particularly at ADJ, as a priority for confirmatory monitoring. They do not establish total metal contents, bioavailability, food-chain exposure or unique pollution sources.

Black soil (Mollisol), characterized by high soil organic matter (SOM), is becoming more vulnerable to extreme heat events, yet the effects of heat on decomposition of SOM, thus N ₂ O emission remain unclear. Here, we investigated the residual effect of temporal heat exposure on subsequent N losses and N ₂ O production pathways in a cultivated black soil in Northeast China. Soil samples were exposed to 25°C, 35°C, 45°C, and 50°C for 12 h, followed by a 7‐day recovery period, after which N ₂ O “hot moment” events (i.e., rewetting with N addition and aerobic‐to‐anaerobic transition) were set up using a robotized continuous‐flow incubation system. We then performed high‐frequency automated monitoring of CO ₂ , N ₂ O, and N ₂ fluxes, and analyzed N ₂ O isotopocules, bacterial community composition, and N‐cycling functional genes to elucidate the underlying mechanisms. Compared with the 25°C control, heat‐exposed soils exhibited 28%–94% higher N ₂ O emissions during the hot moment events. The ¹⁵ N site preference values of N ₂ O indicated that the short‐term heat exposure prior to rewetting enhanced the contribution of denitrification‐derived N ₂ O. The N ₂ O/(N ₂ O + N ₂ ) ratio increased in the extreme heat treatments, indicating a shift toward less complete denitrification. Heating caused modest shifts in bacterial community composition and selectively altered N‐cycling functional gene abundances, including increased AOB amoA and narG abundance in the 50°C treatment. The higher mineral N content and the reduced total N content (by 7%–13%) in heat‐exposed soils after incubation suggest that extreme heat accelerated the decomposition of soil organic N, thereby triggering greater N ₂ O emissions during subsequent hot moment events. The findings have important implications for predicting greenhouse gas feedbacks in black soils under intensifying climate extremes, particularly the potential amplification of soil N ₂ O emissions.

Hyper-arid agricultural soils combine low organic carbon, high calcium carbonate contents, salinity, and frequent gypsum accumulation, making rapid quantification of soil organic carbon (SOC), soil inorganic carbon (SIC), and gypsum valuable but analytically demanding. We evaluated visible–near-infrared (VNIR) spectroscopy for predicting these three properties in 216 samples from 18 plots, nested within six oasis sites spanning two oasis management systems, three topographic positions, and six soil depths, with particular emphasis on how validation design affects apparent model accuracy. Partial least squares regression (PLSR) and random forest (RF), single- and multi-target workflows, and global versus stratified calibrations were compared using plot-blocked GroupKFold, site-blocked GroupKFold, random KFold, and leave-one-out cross-validation (LOOCV). Under continuum-removed spectra, RF outperformed PLSR for SIC, whereas the advantage for SOC and gypsum was not robust to plot or site-level resampling. Predictability was strongly target-specific: gypsum was predicted most reliably (R² = 0.78), SOC moderately (R² = 0.44), and SIC least reliably (R² = 0.30). KFold and LOOCV generally yielded more optimistic estimates than plot-blocked GroupKFold, inflating SIC R² by 0.15–0.20 for the global model and by up to 0.64 for topography-stratified models (pooled across strata). Site-blocked validation further showed that spatial transferability was strongly target-specific: SOC performance remained similar between plot-blocked and site-blocked validation (R² = 0.41 and 0.43, respectively), gypsum performance declined moderately (R² = 0.78 to 0.64), whereas SIC performance decreased markedly from R² = 0.30 to −0.28, indicating poorer performance than a mean-based prediction under oasis-level blocking. VNIR spectroscopy therefore shows strong potential for gypsum, moderate utility for SOC, but limited transferability for SIC. Validation design, together with the spatial scale at which independence is imposed, critically determines how soil-spectroscopy performance should be interpreted for dryland monitoring.

Riparian zones of reservoirs are dynamic hotspots for carbon cycling, yet the response mechanisms of soil organic carbon (SOC) fractions to varying hydrological regimes remain incompletely understood. This study investigated the spatial distribution and driving factors of particulate (POC) and mineral-associated organic carbon (MAOC) across a hydrological gradient in the large subtropical Shanxi Reservoir (watershed area: 1,529 km²), China. Soil samples (Haplic Acrisols) were collected in July 2024 from four zones ranging from permanently inundated to not inundated, with SOC separated into two fractions (POC and MAOC) via particle-size fractionation and analysed using a randomised complete block design in which each of the ten transects spanned all four zones. The results showed that while POC increased more than four-fold from the inundated to the not inundated zone, MAOC exhibited no significant zonal variation. Consequently, the POC/MAOC ratio rose monotonically from 0.3 to 0.8, and inundation zone explained 96% of its variance but only 41% of that in total SOC, demonstrating that inundation frequency regulates carbon compositional structure more strongly than total carbon content. Contrary to conventional mineral protection paradigms, clay and silt contents were negatively correlated with all carbon fractions due to hydrodynamic sorting. Furthermore, carbon composition was governed overall by long-term inundation regimes rather than instantaneous soil moisture states. These findings highlight that the more active and labile carbon pool (i.e. POC) is highly vulnerable to changing hydrological regimes while MAOC as a more stable pool is less affected. The decoupling of clay from carbon stabilisation challenges traditional models in dynamic environments and suggests that reservoir water-level management is an underexplored but critical lever for regulating the stability of riparian carbon pools at the terrestrial-aquatic interface.

Hyper-arid agricultural soils combine low organic carbon, high calcium carbonate contents, salinity, and frequent gypsum accumulation, making rapid quantification of soil organic carbon (SOC), soil inorganic carbon (SIC), and gypsum valuable but analytically demanding. We evaluated visible–near infrared (VNIR) spectroscopy for predicting these three properties in 216 samples from 18 plots, nested within six oasis sites spanning two oasis management systems, three topographic positions, and six soil depths, with particular emphasis on how validation design affects apparent model accuracy. Partial least squares regression (PLSR) and random forest (RF), single- and multi-target workflows, and global versus stratified calibrations were compared using plot-blocked GroupKFold, site-blocked GroupKFold, random KFold, and leave-one-out cross-validation (LOOCV). Under continuum-removed spectra, RF outperformed PLSR across all three targets. Predictability was strongly target-specific: gypsum was predicted most reliably (R² = 0.78), SOC moderately (R² = 0.44), and SIC least reliably (R² = 0.30). KFold and LOOCV generally yielded more optimistic estimates than plot-blocked GroupKFold, inflating SIC R² by 0.15–0.19 for the global model and by up to 0.64 for stratified models. Site-blocked validation further showed that spatial transferability was strongly target-specific: SOC performance remained similar between plot-blocked and site-blocked validation (R² = 0.41 and 0.43, respectively), gypsum performance declined moderately (R² = 0.78 to 0.64), whereas SIC performance decreased markedly from R² = 0.30 to −0.28, indicating poorer performance than a mean-based prediction under oasis-level blocking. VNIR spectroscopy therefore shows strong potential for gypsum, moderate utility for SOC, but limited transferability for SIC. Validation design, together with the spatial scale at which independence is imposed, critically determines how soil-spectroscopy performance should be interpreted for dryland monitoring.