The problem of potentially toxic elements (PTEs) in farmland is a key issue in global pollution prevention and control and has an important impact on environmental safety, human health, and sustainable agricultural development. Based on the climate background of high–latitude cold regions, this study simulated freeze–thaw cycles through indoor tests. Different initial conditions, such as biochar application rates (0%, 1%, 2%) and different initial soil moisture contents (15%, 20%, 25%), were set to explore the morphological changes in cadmium (Cd) and lead (Pb) in soil and the response relationship to the changes in soil physicochemical properties. The results indicate that soil pH decreases during freeze–thaw cycles, and soil alkalinity increases with increasing biochar content. Freeze–thaw cycles caused the total amount of PTEs to have a U–shaped distribution, and the amount of PTEs in the soluble (SOL) and reducible (RED) fraction increased by 0.28–56.19%. Biochar reduced the amount of Cd and Pb migration in the soil, and an increase in soil moisture content reduced the availability of Cd and Pb in the soil. Freezing and thawing damaged the soil structure, and biochar reduced the fractionation of small particle aggregates by enhancing the stability of soil aggregates, thereby reducing the soil's ability to adsorb Cd and Pb. In summary, for farmland soil remediation and pollution control, the application of biochar has a certain ability to optimize soil properties. Considering the distribution of PTEs in the soil and the physicochemical properties of the soil, the application of 1% biochar to soil with a 20% moisture content is optimal for regulating seasonally frozen soil remediation.

In peri-urban critical zones, soil ecosystems are highly affected by increasing urbanization, causing probably an intense interaction between dissolved organic matter (DOM) and heavy metals in soil. Such interaction is critical for understanding the biogeochemical cycles of both organic matter and heavy metals in these zones. However, limited research has reported the correlative distribution of DOM and heavy metals at high seasonal and spatial resolutions in peri-urban critical zones. In this study, 160 soil samples were collected from the farmland and forestland of Zhangxi watershed, in Ningbo, eastern China during spring, summer, fall and winter four seasons. UV–visible absorption and fluorescent spectroscopy were used to explore the optical characteristics of DOM. The results indicated a mixture of exogenous and autogenous sources of DOM in the Zhangxi watershed, while DOM in farmland exhibited a higher degree of aromaticity and humification than that in forestland. Fluorescent results showed that humic acid-like, fulvic acid-like and microbial-derived humic-like fractions were mostly affected by seasons. The distribution of heavy metals was affected mainly by land-use changes and seasons. Correlation analysis between heavy metals and DOM characteristics and components suggested that aromatic and humic substances were more favorable in binding with EDTA extractable Ni, Cu, Zn and Cd. The bioavailable Cd and Pb decreased due to binding with humic fractions, indicating its great effects on the bioavailability of Cd and Pb. Overall, these findings provide an insight into the correlative distributions of DOM and heavy metals in peri-urban areas, thereby highlighting their biogeochemical cycling in the soil environment.

Optimizing landscape pattern to reduce the risk of non-point source (NPS) pollution is an effective measure to improve river water quality. The “source-sink” landscape theory is a recent research tool for landscape pattern analysis that can effectively integrate landscape type, area, spatial location, and topographic features to depict the spatial heterogeneity of NPS pollution. Based on this theory, we quantitatively analyzed the influence of “source-sink” landscape pattern on the river water quality in one of the most intensive agricultural watersheds in Southeastern China. The results indicated that the proportion of “sink” landscape (68.59%) was greater than that of “source” landscape (31.41%) in the study area. In addition, when elevation and slope increased, the “source” landscape proportion decreased, and the “sink” landscape proportion increased. Nitrogen (N) and phosphorus (P) pollutants in rivers showed significant seasonal and spatial variations. Farmland was the primary source of nitrate nitrogen (NO₃⁻-N) and total nitrogen (TN) pollution, whereas residential land was the primary source of ammonium nitrogen (NH₄⁺-N) and total phosphorus (TP) pollution. Intensively cultivated areas and densely inhabited areas degraded water quality despite high proportions of forest land. The four “source-sink” landscape indices (LWLI, LWLI'e, LWLI's, LWLI'd) had significant positive correlations with NO₃⁻-N and TN and weak correlations with NH₄⁺-N and TP. The capacity of LWLI to quantify the NPS pollution was greater in agricultural areas than in residential areas. The “source-sink” landscape thresholds resulted in abrupt changes in water quality. When LWLI was ∼0.35, the probability of river water quality degradation increased sharply. The results suggest the importance of optimizing the “source-sink” landscape pattern for mitigating agricultural NPS pollution and provide policy makers with adequate new information on the agroecosystem-environmental interface in highly developed agricultural watersheds.

Conservation agriculture through no-till based on cropping systems with high biomass-C input, is a strategy to restoring the carbon (C) lost from natural capital by conversion to agricultural land. We hypothesize that cropping systems based on quantity, diversity and frequency of biomass-C input above soil C dynamic equilibrium level can recover the natural capital. The objectives of this study were to: i) assess the C-budget of land use change for two contrasting climatic environments, ii) estimate the C turnover time of the natural capital through no-till cropping systems, and iii) determine the C pathway since soil under native vegetation to no-till cropping systems. In a subtropical and tropical environment, three types of land use were used: a) undisturbed soil under native vegetation as the reference of pristine level; b) degraded soil through continuous tillage; and c) soil under continuous no-till cropping system with high biomass-C input. At the subtropical environment, the soil under continuous tillage caused loss of 25.4 Mg C ha⁻¹ in the 0–40 cm layer over 29 years. Of this, 17 Mg C ha⁻¹ was transferred into the 40–100 cm layers, resulting in the net negative C balance for 0–100 cm layer of 8.4 Mg C ha⁻¹ with an environmental cost of USD 1968 ha⁻¹. The 0.59 Mg C ha⁻¹ yr⁻¹ sequestration rate by no-till cropping system promote the C turnover time (soil and vegetation) of 77 years. For tropical environment, the soil C losses reached 27.0 Mg C ha⁻¹ in the 0–100 cm layer over 8 years, with the environmental cost of USD 6155 ha⁻¹, and the natural capital turnover time through C sequestration rate of 2.15 Mg C ha⁻¹ yr⁻¹ was 49 years. The results indicated that the particulate organic C and mineral associate organic C fractions are the indicators of losses and restoration of C and leading C pathway to recover natural capital through no-till cropping systems.

Cadmium (Cd) is a major pollutant in farmland, which not only greatly restricts crop production, but also brings a serious threat to human health through entering the food chain. Our previous study showed that hemin treatment could reduce the accumulation of Cd in pak choi seedlings. However, the underlying mechanism remains unclear. In this study, we used non-invasive micro-test technology (NMT) to detect the real-time Cd²⁺ flux from pak choi roots and demonstrated that hemin treatment decreased Cd uptake rather than its translocation within plants. Moreover, through comparing the responses of different chemical treatments in pak choi seedlings and Arabidopsis wild-type and heme oxygenase-1 (HO-1) mutant, we provided evidence that hemin-decreased Cd uptake was HO-1 dependent. Furthermore, analyses of hemin degradation products suggested that the hemin-derived suppression of Cd uptake suppression was probably relying on its degradation by-products, ferrous iron (Fe²⁺) and carbon monoxide (CO), via repressing the expression of a Fe²⁺/Cd²⁺ transporter BcIRT1 in pak choi roots.

Forest and agricultural land use affects the concentration and composition of dissolved organic carbon (DOC) in streams and rivers. To elucidate the impacts of forest and agricultural land use on stream DOC during storm events, we investigated DOC concentration ([DOC]), optical properties of dissolved organic matter (DOM), and Δ¹⁴C-DOC in both forest- and agriculture-dominated headwater streams in South Korea in the summer of 2012. One forested and five agricultural streams were investigated. During storms, the peak [DOC] of forest stream increased to 5.8 mg L⁻¹, approximately two times larger than that of the most agricultural stream (3.2 mg L⁻¹), demonstrating the weaker storm responses of the [DOC] of agricultural streams to hydrological change. Five PARAFAC components were identified, including three terrestrial humic-like substances (C1, C2, C3), one microbial humic substance (C4), and one microbial protein-like substances (C5). The mean (C4+C5)/(C1+C2+C3) of all storm events at the most agricultural stream was 1.5 times larger than that of the most forested stream, suggesting that more protein-like DOM is exported from agricultural watersheds. Whereas a forest stream was primarily composed of terrestrially derived and ¹⁴C-enriched modern DOC, the ¹⁴C-age of the most agricultural stream was up to ∼1000 years old. The results suggest that agricultural practices could decrease the old organic carbon pools from soils. However, how quickly the aged DOC can be degraded to CO₂ in streams is unknown, warranting future investigation on lability of the aged DOC and their effects on CO₂ evasion from rivers and estuaries downstream.

The environmental load of organophosphate ester (OPE) flame retardants has caused a series of problems due to their extensive use. The soil matrix, as an ultimate sink for organic pollution, plays a vital part in the fate of OPEs in the environment. In this study, the spatial occurrence, composition profile and health risk of 13 OPE species in farmland soils from four provinces of China were characterized. Excluding tris(2,3-dibromopropyl) phosphate (TDBPP) and ethylhexyl diphenyl phosphate (EHDPP), the remaining eleven OPEs had a high detection frequency (DF) ranging from 60% to 100%. The range of total OPE (ΣOPE) concentrations were 62.3–394 ng/g dry weight (dw), with a median of 228 ng/g dw. Among these OPEs, tris(2-ethylhexyl) phosphate (TEHP) with a median of 143 ng/g dw) was the predominant species, followed by tricresyl phosphate (TCP; median of 20.1 ng/g dw) and tris(2-chloroethyl) phosphate (TCEP; median of 17.9 ng/g dw). In terms of geographical distribution, significantly lower OPEs levels were found in samples from Heilongjiang (159 ± 47.0 ng/g dw) than in those of Guangxi (264 ± 66.0 ng/g dw), Henan (252 ± 74.5 ng/g dw) and Hubei (242 ± 52.8 ng/g dw) provinces. Principal component analysis and Spearman’s correlations were used to reveal potential sources of OPEs in the different provincial regions. Health risk exposure to OPEs in farmland soils was at an acceptable level (<1.20 × 10⁻⁵ for non-carcinogenic risk to children as the most sensitive age group; and <6.47 × 10⁻¹⁰ for carcinogenic risk to adults as the most sensitive age group) at the present detected concentrations. However, TCEP and TEHP, the predominant risk contributors, should be paid more attention.

Mapping soil contamination enables the delineation of areas where protection measures are needed. Traditional soil sampling on a grid pattern followed by chemical analysis and geostatistical interpolation methods (GIMs), such as Kriging interpolation, can be costly, slow and not well-suited to highly heterogeneous soil environments. Here we propose a novel method to map soil contamination by combining high-resolution aerial imaging (HRAI) with machine learning algorithms. To support model establishment and validation, 1068 soil samples were collected from an arsenic (As) contaminated area in Zhongxiang, Hubei province, China. The average arsenic concentration was 39.88 mg/kg (SD = 213.70 mg/kg), with individual sample points determined as low risk (66.9%), medium risk (29.4%), or high risk (3.7%), respectively. Then, identified features were extracted from a HRAI image of the study area. Four machine learning algorithms were developed to predict As risk levels, including (i) support vector machine (SVM), (ii) multi-layer perceptron (MLP), (iii) random forest (RF), and (iii) extreme random forest (ERF). Among these, we found that the ERF algorithm performed best overall and that its prediction performance was generally better than that of traditional Kriging interpolation. The accuracy of ERF in test area 1 reached 0.87, performing better than RF (0.81), MLP (0.78) and SVM (0.77). The F1-score of ERF for discerning high-risk points in test area 1 was as high as 0.8. The complexity of the distribution of points with different risk levels was a decisive factor in model prediction ability. Identified features in the study area associated with fertilizer factories had the most important contribution to the ERF model. This study demonstrates that HRAI combined with machine learning has good potential to predict As soil risk levels.

Controlling ammonia (NH₃) emissions has been proposed as a strategy to mitigate haze pollution. To explore the role of NH₃ in haze pollution in Sichuan Basin, where agricultural activities are intense, hourly in situ data of NH₃, as well as nitric acid and secondary inorganic aerosols (SIAs) were gathered in Chengdu from April 2017 to March 2018. We found that NH₃ had an annual mean concentration of 9.7 ± 3.5 (mean ± standard deviation) μg m⁻³, and exhibited seasonal variations (spring > summer > autumn and winter) due to changes in emission sources and meteorological conditions (particularly temperature). Chengdu's atmosphere is generally NH₃-sufficient, especially in the warm seasons, implying that the formation of SIAs is more sensitive to the availability of nitric acid. However, an NH₃ “sufficient-to-deficient” transition was found to occur during winter pollution periods, and the frequency of NH₃ deficiency increased with the aggravation of pollution. Under NH₃-deficient conditions, the nitrogen oxidation ratio increased linearly with the increase in free NH₃, implying that NH₃ contributes appreciably to the formation of nitrate and thus to high PM₂.₅ loadings. No relationships of NH₃ with fossil fuel combustion–related pollutants were found. The NH₃ emissions from farmland and livestock waste in the suburbs of Chengdu and regional transport from west of Chengdu probably contribute to the occurrence of high PM₂.₅ loading in winter and spring, respectively. These results suggest that to achieve effective mitigation of PM₂.₅ in Chengdu, local and regional emission control of NH₃ and NOx synergistically would be effective.

There must be some uncertainty in the remediation areas delineated based on limited sample points, and resampling in the high-uncertainty areas is particularly necessary. In situ field portable X-ray fluorescence spectrometry (FPXRF), a rapid and cheap analysis method for soil heavy metals, is strongly affected by many spatially non-stationary soil factors. This study first delineated the high-uncertainty area (threshold-exceeding probabilities (PTE) between 30% and 70%) of soil Pb based on the 1000 realizations produced by sequential Gaussian simulation (SGS) with 93 ICP-MS Pb concentrations measured in a peri-urban agriculture area, China. Next, in situ FPXRF was used to increase sample density in this high-uncertainty area. Then, robust geographically weighted regression (RGWR) was used to correct the in situ FPXRF Pb, and the correction accuracies of RGWR, basic GWR, and traditionally-used ordinary least squares regression (OLSR) were compared. Finally, to explore the best way to combine these corrected in situ FPXRF concentrations in delineating the remediation area, we compared the following spatial simulation methods: basic SGS, sequential Gaussian co-simulation (CoSGS) with the RGWR-corrected in situ FPXRF Pb as auxiliary soft data (CoSGS-CorFPXRF), and SGS with the RGWR-corrected in situ FPXRF Pb as part of hard data (SGS-CorFPXRF). Results showed that (i) RGWR produced higher correction accuracy (RI = 71.5%) than GWR (RI = 59.68%) and OLSR (RI = 25.58%) for the in situ FPXRF Pb; (ii) SGS-CorFPXRF produced less uncertainty (G = 0.97) than CoSGS-CorFPXRF (G = 0.95) and SGS (G = 0.91) in the spatial simulation; (iii) High-uncertainty area (30%<PTE<70%) was reduced from 36.55% to 8.7% of the whole study area. It is concluded that the recommended methods are cost-effective to reduce the uncertainty in delineating the remediation areas of soil heavy metals.