The change in leaf color during the later reproductive period of rice is directly related to photoassimilate accumulation and nutrient reuse, and it ultimately affects grain filling and yield.  This study aimed to explore an assessment model that depicts the leaf color change process, and extract parameters that can precisely distinguish differences in leaf color changes among different treatments and varieties.  A total of 31 rice varieties were selected as the field experiment materials in 2019 and 2023.  The SPAD values of the flag, 2nd and 3rd leaves were measured after heading, and they were normalized to the leaf color index (CI).  A functional model for the variation of leaf CI with time (t) in the late reproductive stage of rice was established based on CI=at²+bt+c, and seven color change parameters were extracted for the quantitative comparison and assessment of leaf color changes, including three time related parameters for color change (onset time, T₀; midpoint time, T₅₀; and color change duration, T₁₀₀); one leaf color index (final value of CI, CI_f); and three parameters related to the color change rate (the rate during T₀−T₅₀, R₁; the rate during T₅₀−T₁₀₀, R₂; and the mean color change rate, R_m).  In 2023, Chunyou 927 (CY927) with a dark leaf color and Yongyou 1540 (YY1540) with a normal leaf color were used as materials, and three N fertilizer amounts were applied to explore the effects of N fertilizer on the leaf color change process through the established assessment system.  The T₀ of the flag leaf was delayed by 2.6−3.0 d compared to the 2nd and 3rd leaves.  The CI_f of the flag leaf was 12.12 and 21.15% higher than those of 2nd and 3rd leaves, respectively.  In addition, the R₁, R₂ and R_m of the 3rd leaf were 10.75–19.82%, 17.99–20.09% and 18.23–11.61% higher than the flag and 2nd leaves, respectively.  Rice yield was significantly positively correlated with T₀, positively correlated with T₅₀ and T₁₀₀, and negatively correlated with R₁, R₂ and R_m.  The average T₀, T₅₀, and T₁₀₀ of rice varieties with yields higher than 8,000 kg ha⁻¹ were 6.8, 22.2, and 31.8 d, respectively, with a CI_f of 0.563 and an R_m of 0.015 d^–1.  N applications delayed T₀ by 4.5–6.2 d, reduced R_m by 30.06–32.33%, and increased CI_f by 35.78–39.69%.  The established leaf color change model and extracted parameters quantitatively depicted the leaf color change process during the later reproductive period.  They also effectively distinguished the differences in leaf color change among leaf positions, rice varieties and N treatments.  This approach is valuable for selecting and cultivating high-yield and nutrient-efficient rice varieties, as well as for analyzing the underlying mechanisms.

Flavonols and flavanones are important bioactive compounds with multiple pharmacological activities and health benefits.  Transcriptional activation of flavonol and flavanone biosynthesis has been studied extensively, while little is known about the negative regulators.  CRISPR/Cas9 gene-editing technology, with the advantage of precise genetic modification, is a desirable tool for breeding biofortified materials and exploring potential molecular mechanisms.  In this study, a transcriptional repressor, SlMYB32, was characterized in tomato fruit.  Phenotype and metabolomic analyses confirmed that knockout of SlMYB32 resulted in increased accumulation of flavonols and flavanones, especially about 1 mg g^–1 FW of quercetin 3-O-rutinoside (rutin).  Transcriptome analysis indicated that expression of key genes SlPAL6, Sl4CL3 and Sl4CL4 as well as five candidate SlUGTs were significantly up-regulated in slmyb32 mutants.  Dual-luciferase and EMSA assays indicated SlMYB32 could bind to and repress promoter activities of SlPAL6 and Sl4CL3.  Expression of 27 transcription factors belonging to 12 families was significantly changed in slmyb32 mutants, among which two SlMYBs, two SlNACs, two SlAP2s and one SlWRKY were clustered with known flavonoid regulators.  Our results provide new insights into improving bioactive compounds in fruit and understanding negative regulatory mechanisms in flavonol and flavanone biosynthesis.

Understanding the spatial distributions and corresponding variation mechanisms of key soil nutrients in fragile karst ecosystems can assist in promoting sustainable development.  However, due to the implementation of ecological restoration initiatives such as land-use conversions, novel changes in the spatial characteristics of soil nutrients remain unknown.  To address this gap, we explored nutrient variations and the drivers of the variation in the 0–15 cm topsoil layer using a regional-scale sampling method in a typical karst area in northwest Guangxi Zhuang Autonomous Region, Southwest China.  Descriptive statistics, geostatistics, and spatial analysis were used to assess the soil nutrient variability.  The results indicated that soil organic carbon (SOC), total nitrogen (TN), total phosphorus (TP), and total potassium (TK) concentrations showed moderate variations, with coefficients of variance being 0.60, 0.60, 0.71, and 0.72, respectively.  Moreover, they demonstrated positive spatial autocorrelations, with global Moran’s indices being 0.68, 0.77, 0.64, and 0.68, respectively.  However, local Moran’s index values were low, indicating large spatial variations in soil nutrients.  The best-fitting semi-variogram models for SOC, TN, TP, and TK concentrations were spherical, Gaussian, exponential, and exponential, respectively.  According to the classification criteria of the Second National Soil Census in China, SOC and TN concentrations were relatively sufficient, with the proportions of rich and very rich levels being up to 90.9 and 96.0%, respectively.  TP concentration was in the medium-deficient level, with the areas of medium and deficient levels accounting for 33.7 and 30.1% of the total, respectively.  TK concentration was deficient, with the cumulative area of extremely deficient, very deficient, and deficient levels accounting for 87.6% of the total area.  Consequently, the terrestrial ecosystems in the study area were more vulnerable to soil P and K than soil N deficiencies.  Furthermore, variance partitioning analysis of the influencing factors showed that, except for the interactions, the single effect of other soil properties accounted more for soil nutrient variations than spatial and environmental variables.  These results will aid in the future management of terrestrial ecosystems.

The United Nations Sustainable Development Goal (SDG) 2 aims to achieve Zero Hunger by 2030.  However, global hunger and food insecurity have continued to rise at an alarming rate (UN 2023).  Subtropical regions are home to more than 30% of the world’s population, predominantly in developing countries where per capita farmland and food supply are only 40% of those in developed nations (FAO 2018).  Meeting the Zero Hunger target amid ongoing population growth in these regions requires a substantial increase in agricultural production while minimizing soil degradation and adverse ecological impacts.  This challenge is shared by many countries across South Asia, Africa, and Central and South America.

Against this background, the 4th International Symposium on Sustainable Agriculture for Subtropical Regions (ISSASR-4) was held from June 21 to 24, 2024, in Changsha, China, hosted by the Institute of Subtropical Agriculture, Chinese Academy of Sciences.  The symposium brought together over 300 experts and scholars from nearly 30 countries.  Under the theme “Ecosystem Management and Agricultural Green Development in Subtropical Regions”, discussions focused on four key topics: (i) regional resources and ecosystem management for enhancing agricultural productivity, (ii) green crop and animal production, (iii) minimizing adverse environmental impacts of agricultural production, and (iv) the growing role of big data, artificial intelligence (AI), and smart farming.  Participants exchanged the latest research advances, identified major challenges, and explored countermeasures for agriculture and ecological sustainability in subtropical regions worldwide.

This Special Focus of the Journal of Integrative Agriculture (JIA) addresses these pressing issues by presenting empirical evidence and innovative solutions for agricultural green development.  It comprises 13 papers covering a wide range of topics related to carbon, nitrogen, and phosphorus pathways in natural and agricultural ecosystems, with attention to microbial processes, land-use change, production management, and their effects on nutrient cycling and grain yield.  We hope this collection enhances understanding of ecosystem management and green agricultural development, offering actionable insights for policymakers, researchers, and practitioners.

Section 1: Regional resources and ecosystem management

This section examines three key areas: agricultural bio-resources, soil carbon and nutrient dynamics across ecosystems, and regional grain supply–demand matching.  Studies provide insights into bioinput-based agricultural frameworks, soil nutrient responses to climate change and anthropogenic influences, and the dynamic, heterogeneous patterns of grain matching.  Vermelho et al. (2026) reviewed microbial bioinputs, outlining their categories, mechanisms, global challenges, and Brazil’s production infrastructure and regulatory context.  Wang M M et al. (2026) reported moderate spatial variation with positive autocorrelation in soil organic carbon (SOC), total nitrogen (TN), total phosphorus (TP), and total potassium (TK), noting greater vulnerability to phosphorus and potassium limitation than to nitrogen, with soil properties outweighing spatial or environmental factors in explaining nutrient variation.  Another study by Wang L Y et al. (2026) identified climate and hydrological changes as key drivers of SOC loss in Dongting Lake, with accelerated loss occurring above 21.4 m elevation, suggesting that managed water levels during droughts could enhance carbon sequestration.  Wan et al. (2026) showed that plantations can mitigate climate change by increasing carbon storage at the aggregate scale in alpine regions.  Miao et al. (2026) demonstrated a scale-dependent mismatch in grain supply and demand, highlighting how interregional flows from 1980 to 2020 reduced deficit areas.  Together, these studies advance frameworks for sustainable ecosystem management.

Section 2: Green crop production in subtropical regions

Enhancing green crop production in subtropical regions requires practices that improve soil health and carbon sequestration while sustaining yields.  Given the vulnerability of subtropical croplands, effective strategies for maintaining SOC are critical.  Hua et al. (2026) found that long-term livestock manure substitution improves soil aggregate stability and reduces water erosion but increases lateral loss of labile organic carbon, revealing a trade-off.  Kautsar et al. (2026) reported that terrace reconstruction altered rice yields between field sides and modified SOC, TN, and decomposition dynamics in the 15–30 cm layer, with subsoil fertility determining productivity.  Wang J et al. (2026) demonstrated that massive granulated straw incorporation boosts SOC and crop yield in infertile soils, with accumulation efficiency ranging from 30.8 to 60.0%, primarily from plant residues.  These studies highlight practical pathways for sustainable soil management.

Section 3: Environmental impacts of agricultural production

Assessing and mitigating agriculture’s environmental footprint requires a multiscale understanding of soil ecological processes.  Pan et al. (2026) found that natural restoration enhances karst soil phosphorus-cycle multifunctionality more than artificial restoration or cropping, driven by increased SOC and bacterial network complexity, with rare phoD-harboring taxa playing a critical role.  Wang Y et al. (2026) reported that niche outweighs genotype in shaping pea fungal communities, with β-diversity driven by species replacement and deterministic assembly in niche-based groups.  Zhu et al. (2026) showed that SOC is higher in brown and yellow-brown soils and that spring irrigation significantly increases farmland SOC, supporting carbon sequestration.  Zheng et al. (2026) demonstrated that spatial factors govern carbon-cycling gene abundances in uplands, while biotic and substrate factors dominate in paddy soils, revealing an integrated “microbial carbon pump” in trace-gas cycling at a continental scale.  Collectively, these studies advance understanding of the mechanisms underlying soil functionality and greenhouse gas modulation.

Section 4: Big data, artificial intelligence and smart farming in agriculture

The integration of big data, AI, and smart technologies is pivotal for the digital transformation of agriculture.  This section presents a study on their practical application to environmental challenges.  Wang M H et al. (2026) developed an Android-based decision support system (CNPDSS) to control non-point source nitrogen (N) and phosphorus (P) pollution.  Integrating GIS, a Bayesian predictive model, an optimization algorithm, and a smartphone interface, the system identified solutions that minimize both pollutant loadings and engineering costs in the Tuojia catchment, China.  Its adaptive design demonstrates potential for broader application, offering a scalable tool for sustainable water quality management.

This Special Focus underscores the critical intersection of ecosystem management and agricultural development in subtropical regions.  Through 13 studies organized across four themes - resource management, green production, environmental impact mitigation, and smart technology - the collection provides a science-based framework for enhancing productivity while preserving ecological integrity.  It offers concrete insights for achieving sustainable food systems and advancing the UN Zero Hunger goal in some of the world’s most vulnerable and vital agricultural landscapes.

Nitrogen (N) and potassium (K) are two key mineral nutrient elements involved in rice growth.  Accurate diagnosis of N and K status is very important for the rational application of fertilizers at a specific rice growth stage.  Therefore, we propose a hybrid model for diagnosing rice nutrient levels at the early panicle initiation stage (EPIS), which combines a convolutional neural network (CNN) with an attention mechanism and a long short-term memory network (LSTM).  The model was validated on a large set of sequential images collected by an unmanned aerial vehicle (UAV) from rice canopies at different growth stages during a two-year experiment.  Compared with VGG16, AlexNet, GoogleNet, DenseNet, and inceptionV3, ResNet101 combined with LSTM obtained the highest average accuracy of 83.81% on the dataset of Huanghuazhan (HHZ, an indica cultivar).  When tested on the datasets of HHZ and Xiushui 134 (XS134, a japonica rice variety) in 2021, the ResNet101-LSTM model enhanced with the squeeze-and-excitation (SE) block achieved the highest accuracies of 85.38 and 88.38%, respectively.  Through the cross-dataset method, the average accuracies on the HHZ and XS134 datasets tested in 2022 were 81.25 and 82.50%, respectively, showing a good generalization.  Our proposed model works with the dynamic information of different rice growth stages and can efficiently diagnose different rice nutrient status levels at EPIS, which are helpful for making practical decisions regarding rational fertilization treatments at the panicle initiation stage.

Retrotransposons, a type of DNA fragment that can mobilize itself on genome, can generate genetic variations and develop for molecular markers based on the insertion polymorphism.  Zinc finger proteins (ZNFs) are among the most abundant proteins in eukaryotic animals, and their functions are extraordinarily diverse and particularly important in gene regulation.  In the current study, bioinformatic prediction was performed to screen for retrotransposon insertion polymorphisms (RIPs) in six ZNF genes (ZNF2, ZNF3, ZNF7, ZNF8, ZNF10 and ZNF12).  Six RIPs in these ZNFs, including one short interspersed nuclear element (SINE) RIP in intron 1 and one long interspersed nuclear element 1 (L1) RIP in intron 3 of ZNF2, one SINE RIP in 5´ flanking region and one SINE RIP in intron 2 of ZNF3, one SINE RIP in 3´ UTR of ZNF7 and one L1 RIP in intron 2 of ZNF12, were discovered and their presence was confirmed by PCR.  The impact of the SINE RIP in the first intron of ZNF2, which is close to the core promoter of ZNF2, on the gene activity was investigated by dual-luciferase assay in three cell lines.  Our results showed that the SINE insertion in the intron 1 of ZNF2 repressed the core promoter activity extremely significantly (P<0.01) in cervical cancer cells and porcine primary embryonic fibroblasts (HeLa and PEF), thus SINE may act as a repressor.  This SINE RIP also significantly (P<0.05) affected the corrected back fat thickness in Yorkshire pigs.  The corrected back fat thickness of individuals with SINE insertion in the first intron of ZNF2 was significantly (P<0.05) higher than that of individuals without SINE insertion.  In summary, our data suggested that RIPs play important roles in the genetic variations of these ZNF genes and SINE RIP in the intron 1 of ZNF2 may provide a useful molecular marker for the screening of fat deposition in the pig breeding.

To elucidate the relationship between leaf color-changing and stem NSC translocation during grain filling and their impact on yield formation, two indica-japonica hybrid varieties with distinct leaf color change patterns were planted under three N fertilizer dosages (LN 0 kg ha⁻¹; MN 150 kg ha⁻¹; HN 300 kg ha⁻¹). Leaf color change characteristics, photosynthetic productivity, stem NSC translocation, yield and harvest index were analyzed. The results showed that CY927 (slow leaf color change) achieved 10.45%−21.81% higher yields than YY1540 (fast leaf color change) under high-temperature conditions. Compared to YY1540, CY927 delayed the onset of leaf color-changing (T₀) by 2.1−4.1 d, enhanced the final leaf color indicator at maturation (CI_f) by 16.79−52.25%, contributing to 10.56−42.77% greater aboveground biomass accumulation through higher photosynthetic capacity, but significantly limited stem NSC remobilization, reduced total NSC translocation by 23.78−33.19% and NSC translocation ratio by 14.65−22.19%, resulting in a 2.66−8.43% lower harvest index. N application increased rice yield via a delay in leaf color-changing onset (T₀), a reduced color-changing rate (R_m), a shortened color-changing duration (T₁₀₀), and an improved final color index (CI_f). This retardation of senescence enhanced photosynthetic capacity, which was associated with elevated sucrose content and sucrose synthase activity. However, N reduced stem α-amylase activity (14.83−62.07%) and NSC translocation ratio (5.44−16.30%) in both varieties. Correlation analysis revealed significant positive relationships between T₀ and aboveground biomass (P<0.001), and between T₁₀₀ and stem NSC translocation (P<0.001). In conclusion, rice variety and N application indirectly regulate the dynamic balances between leaf photosynthetic carbon metabolism and stem NSC translocation by influencing the leaf color-changing dynamic, ultimately affecting yield and resource use efficiency. This integrative framework, connecting leaf color-changing, carbon allocation, and yield performance, provides scientific guidance for optimizing rice cultivars and N fertilization strategies.

Accurate estimation of Leaf Area Index (LAI) in multi-variety rice using optical remote sensing remains challenging due to spectral saturation under dense canopy conditions and inter-varietal physiological differences. To address this, we developed a multimodal data fusion framework integrating RGB and multispectral imagery acquired by unmanned aerial vehicles (UAVs), combined with features derived from Digital Surface Models (DSM), vegetation indices (VIs), texture, and depth representations. Using field data collected across 60 rice varieties, four machine learning models were evaluated for LAI estimation. Our results demonstrate that multimodal fusion substantially outperforms conventional VI-based approaches. Among them, the Random Forest Regression (RFR) model achieved optimal performance (R²=0.76, RMSE=0.57), representing a 26–58% improvement in R² over baseline models. SHAP-based feature importance analysis identified DSM feature, height-stratified vegetation indices, and depth features as key contributors to model accuracy. This study establishes that incorporating canopy structural information and deep features mitigates saturation effects and enhances generalizability across varieties. The proposed approach offers a robust and efficient solution for high-throughput LAI estimation, supporting applications in precision agriculture and rice breeding programs.

This study explored the complex mechanisms of methane (CH₄) emissions in paddy fields, focusing on the often-overlooked role of soil texture. Through the analysis of 31 paddy soil samples, the research investigated the complex interactions among soil texture, organic carbon composition, soil nutrients, and microbial abundance in regulating CH₄ emissions during the tillering stage of rice. The results revealed significant variations in CH₄ emissions among different soils, which were notably associated with soil texture, organic carbon, nutrients levels, and microbial abundance. Soil texture, particularly clay content, emerged as a key factor influencing the composition of organic carbon, showing a significant positive correlation with mineral-associated organic carbon (MAOC). While organic carbon components significantly enhanced CH₄ emissions, their effects were not uniform: particulate organic carbon correlated negatively with emissions, whereas MAOC showed a positive association. Soil texture also influenced nutrients availability, with clay content significantly correlated with soil nitrogen and phosphorus content, which in turn affects the abundance of functional genes. Specifically, mcrA abundance was positively correlated with available potassium, while pmoA abundance was positively correlated with available phosphorus. Additionally, dissolved organic carbon promoted pmoA abundance, although this effect was mitigated by higher clay content. Network analysis further emphasized the central role of soil texture, with clay exhibiting the highest degree and closeness centrality. In conclusion, soil texture is a fundamental and core factor influencing CH₄ emissions at tillering of rice, exerting its influence through multiple pathways including modulating the composition of organic carbon, nutrient availability, and the abundance of methanogens and methanotrophs. These findings provide theoretical foundations for developing low-carbon cultivation strategies tailored to different soil textural characteristics.