Sichuan is located in southwestern China in the upper reaches of the Yangtze River (Figure 1), with a regional range of 97°21′E–108°31′E and 26°3′N–34°19′N. The topographical structure is ‘high around the edges, low in the middle’, with the central area characterized by typical basin terrain. This region has a subtropical humid monsoon climate, with significant monsoon and vertical zonal characteristics. Annual average temperatures range from 16 to 18°C, while annual precipitation varies between 800 and 1,200 mL, distributed uniformly across space and time. The frost-free period extends for 240–300 days, with abundant light and heat and sufficient heat accumulation, providing favorable climatic conditions for crop growth. Soil types primarily consist of paddy and purple soil, with deep soil layers and high organic matter content, offering good water and fertilizer retention. These soil characteristics have established a necessary foundation for achieving high and stable grain crop yields. This region is not only one of China’s important commercial grain production areas but also a crucial ecological barrier in the upper reaches of the Yangtze River.

From 2000 to 2023, Sichuan Province’s economic and agricultural development shows distinct phases and notable regional variations (Zhang et al., 2024). The agricultural planting industry exhibits clear regional development disparities, with leading modernization levels but facing “high input-low efficiency” issues, while the central plains face considerable ecological pressure. Since 2016, the province has entered a period of structural adjustment, shifting its development focus from quantity growth to quality and efficiency improvement (Liu et al., 2024; Xiang et al., 2024). This indicates that Sichuan is transforming from an agricultural big province to an agricultural strong province, though it still faces challenges related to grain production and demand gaps. Coordinated regional development and green transformation will be crucial for future development.

The primary data sources include the Sichuan Provincial Bureau of Statistics (2023), as well as local statistical yearbooks from cities and prefectures in Sichuan, such as the Chengdu Municipal Bureau of Statistics (2023), Zigong Municipal Bureau of Statistics (2023), Sichuan Climate Resources, and Sichuan Provincial Profile. The details are provided in Supplementary Tables S1–S5. To address missing data in the yearbooks, we supplemented them using statistical data published on the official websites of local agricultural bureaus, meteorological bureaus, water affairs bureaus, and other institutions. We performed sensitivity analyses on the data to assess the robustness and quantify uncertainty. Additionally, this research used an emergy baseline of 1.583 × 10²⁵ sej/a (Odum and Odum, 2000), while the emergy conversion coefficients and transformation rates were based on Wu (Wu, 2016). Refer to Table 3 for indicators used in emergy analysis evaluation. The details of conversion rate are provided in Supplementary Table S6. The emergy analysis research framework is shown in Figure 4.

A high emergy input density indicates greater energy investment per unit of farmland, although excessive inputs may result in resource wastage and significant environmental impacts. As illustrated in Figure 5A, our analysis showed that regions with high emergy input density were concentrated in central Sichuan. Chengdu had the highest density (10.59 × 10¹¹ sej·m⁻²), while Garzê showed the lowest (5.36 × 10¹¹ sej·m⁻²), which was merely half of Chengdu’s level. The Chengdu Plain offers flat terrain ideal for large-scale mechanization, resulting in high emergy intensity levels. In contrast, the complex western terrain of Garzê severely limits mechanical farming. Our statistical analysis (Figure 5B) revealed that data dispersion was large in high-level and medium-high-level regions, suggesting diversity and complexity in emergy input practices. Meanwhile, the medium-low and low-level regions displayed more concentrated data patterns. We observed a significant overlap between adjacent categories, particularly between high and medium-high levels, as well as between medium-low and low levels, indicating gradual transitions in regional emergy input density. One-way analysis of variance confirmed significant differences among the four groups (F = 120.6, p < 0.001). The large effect size (η² = 0.96) suggests that regional grouping explains a substantial portion of emergy input density variations.

The emergy output density reflects the output benefits of crop production, with higher values indicating a higher emergy output per unit area, that is, higher returns. Our spatial analysis, as illustrated in Figure 5C, reveals that high-level and medium-high-level regions are concentrated in central and eastern Sichuan, whereas medium-low and low-level regions were mainly found in western Sichuan. Aba showed the lowest density at 5.16 × 10¹¹ sej·m⁻², representing only 40% of Dazhou’s output. As shown in Figure 5D, from high-level to low-level regions, the emergy output density showed a clear decreasing trend, potentially reflecting more diverse economic structures in these areas. We observed minor differences between the medium-high and medium-low level regions. One-way analysis of variance confirmed significant differences between groups (F = 71.45, p < 0.001), with an effect size (η² = 0.94) indicating that regional grouping accounts for a substantial portion of variations in emergy output density.

A higher emergy investment ratio indicates a more advanced level of economic development, whereas a lower ratio suggests less development and greater environmental dependence. As illustrated in Figure 6A, Chengdu showed the highest emergy investment ratio at 4.13, primarily because of its substantial input of purchased auxiliary emergy relative to the total emergy input. In contrast, Garzê had the lowest ratio of 1.67, just 40% of Chengdu’s value, mainly because of its low proportion of renewable organic emergy input to total emergy. Regions with high development levels typically demonstrate superior emergy investment efficiency, likely because of their advanced technology, well-developed infrastructure, and effective resource allocation systems (Figure 6B). However, significant variations within these high-level regions suggest room for further optimization. Medium-low level regions showed the least internal variation, while low-level regions displayed greater differences, suggesting that medium-low level regions have more consistent energy utilization patterns, whereas low-level regions showed more variability and potential for improvement. Statistical analysis using one-way analysis of variance revealed significant differences between the four groups (F = 168.37, p < 0.001), with an effect size (η² = 0.97) indicating that group classification strongly influences observed variations.

The net emergy yield ratio is a standard measure of system productivity, with higher values reflecting greater system production efficiency. As shown in Figure 6C, only two regions, Deyang and Chengdu, were in the low-level category, with Chengdu showing the lowest net emergy yield ratio at 1.17, indicating a highly-level lack of coordination between crop production and economic development. Among the high-level regions, Ziyang demonstrated the highest net emergy yield ratio at 2.07, representing a 76% increase over Chengdu’s value. Both high- and low-level regions showed small data fluctuations, with high-level regions displaying more concentrated data distribution, suggesting that these areas may have achieved stable energy efficiency (Figure 6D). In contrast, medium-high and medium-low level regions indicated more scattered data patterns, potentially indicating that they are at a crucial stage of improving energy efficiency, with substantial potential for improvement. Statistical analysis using one-way analysis of variance revealed significant differences among the four groups (F = 455.7, p < 0.001), with an effect size (η² = 0.98) demonstrating that group differences account for almost all net emergy yield ratio variations.

Labor input is the primary factor that influences the scale management degree. Within the same crop production system, a higher labor input typically correlates with a lower scale management degree. Additionally, labor requirements are strongly correlated with farmland topography, with flat terrain facilitating large-scale farming. In Figure 7A, low-level regions included Garzê and Aba, averaging 21.64, primarily due to their hilly terrain, uneven farmland, and heavy reliance on manual farming methods. High-level regions averaged 57, with Suining showing the highest scale management degree at 59.8, largely because of its central location in the Sichuan Basin, with flat terrain and concentrated corn cultivation that enables mechanized production and large-scale management. High-level regions consistently demonstrated elevated and uniform scale management degrees, while medium-high level regions showed more dispersed data distribution. As the level decreases, both the overall scale of management degree declines and regional differences widen (Figure 7B). Statistical analysis using one-way analysis of variance revealed significant differences between the four groups (F = 66.89, p < 0.001), with an effect size (η² = 0.92) indicating that group classification strongly influences variations in the scale management degree.

Excessive use of non-renewable resources by humans can lead to severe ecological degradation, with environmental loading increasing as the consumption of non-renewable resources increases. As shown in Figure 7C, low-level and medium-low level regions have environmental loading ratios below 1, accounting for 67% of Sichuan, indicating that corn cultivation in these cities did not significantly impact the environment. The average environmental loading ratio is 1.29 for medium-high level regions and 1.71 for high-level regions, suggesting that it is important to emphasize non-renewable resource inputs in corn planting systems within these areas. High-level regions showed concentrated data, with a narrow distribution range (Figure 7D). The medium-high level regions demonstrated the greatest data dispersion, with the widest distribution range. In contrast, the medium-low and low-level regions maintained lower and stable environmental loading ratios. One-way analysis of variance revealed significant differences between the four groups (F = 185.96, p < 0.0001), with an effect size (η² = 0.97) indicating that group classification profoundly impacts result variation, showing substantial and significant differences across all levels.

To identify the optimal trend surface model, we spatially applied first-, second-, third-, fourth-, and fifth-order trend surface models to simulate the environmental sustainability index of Sichuan’s corn planting ecosystem (Figure 8A). The first-order trend surface model demonstrated the highest goodness of fit (R² = 0.63). Consequently, this model was used to evaluate spatial variation in the environmental sustainability index of Sichuan’s corn planting ecosystem. It represents patterns that emerge after eliminating random and local fluctuations. In Figure 8B, despite local undulations and fluctuations in central Sichuan, the overall environmental sustainability index showed a decreasing gradient from northeast to southwest, displaying an inclined plane-shaped spatial distribution pattern. This model’s parameters were statistically significant (F = 121.15, p < 0.05) and the regression model exhibited a high goodness of fit for the cross-sectional data (R² > 0.5).

According to the loading plot (Figure 9A), the goodness of fit (R² = 0.89) demonstrated that the PLS-DA model effectively differentiated the environmental sustainability index of corn planting ecosystems across the province. The 21 prefecture-level cities/autonomous prefectures were distinctly categorized into four groups: Group 1 (YA, CD, DY, PZ, and MS), Group 2 (GA, NJ, SN, DZ, and ZY), Group 3 (GY, BZ, NC, LZ, ZG, YB, LS, and SN), and Group 4 (AB, LSYZ, and GZ). Group 1 showed a strong correlation with emergy input density, emergy investment ratio, and environmental loading ratio, indicating that this group exhibited higher values in these metrics. Group 2 correlated strongly with emergy output density, net emergy yield ratio, and scale management degree, suggesting that this group demonstrated higher performance in these aspects. Group 3 showed a weak correlation with all emergy indicators, including the emergy investment ratio, environmental loading ratio, emergy input/output densities, scale management degree, and net emergy yield ratio, exhibiting a greater distance from these metrics. Group 4 was the most distant from these emergy indicators, and showed the weakest correlation. VIP is an indicator for measuring the importance of explanatory variables, with values exceeding 1, typically indicating a significant influence on the response variables. In Figure 9B, the net emergy yield ratio, emergy output density, and scale management degree all show VIP values greater than 1, indicating that variations in these indicators contribute to significant spatial differences in corn sustainability across Sichuan.

This study uses an emergy analysis methodology to evaluate the sustainable development of corn planting systems across 21 prefecture-level cities/autonomous prefectures in Sichuan. In terms of emergy input and output, Dazhou exhibits low emergy input density but high emergy output density, indicating efficient resource utilization with high yields, despite low resource investment per unit area. Located in eastern Sichuan’s basin’s peripheral mountains with a well-developed environment, Dazhou benefits from favorable ecological conditions and optimal planting density, resulting in high corn productivity. Regarding emergy investment and net yield, Chengdu and Deyang had high emergy investment ratios but low net emergy yield ratios, reflecting advanced economic development with reduced environmental dependence and low production efficiency, indicating a significant imbalance between crop production and economic development. In terms of scale management degree and environmental loading ratio, Suining and Ziyang demonstrated high scale management with low environmental loading ratios, characterized by minimal labor input, efficient management, and reduced environmental impact, making them ideal regions for corn cultivation. Conversely, Aba and Ya’an show a low scale management degree but high environmental loading ratios, indicating limited mechanization, inefficient management, and significant environmental pressure, necessitating adjustments in resource utilization, particularly non-renewable resources. By 2025, the grain production and consumption gaps are projected to reach 14.7343 million tons, requiring an additional 2,644.58 thousand hectares of cultivation area, based on Sichuan’s current grain yield. Amidst resource constraints, meeting the growing demand for diverse, high-quality grain remains the primary challenge for Sichuan’s food security (Yusha, 2021).

Natural emergy inputs to corn production systems (including solar energy, wind energy, and precipitation) remained regionally stable, providing baseline energy support for system operations. However, significant regional variations exist in purchased emergy inputs such as fertilizers, pesticides, and machinery. These human energy interventions create distinct ecological and economic characteristics for the same crop across different regions (Asgharipour et al., 2019; Nan et al., 2020). For example, corn production systems in the Chengdu Plain (including Chengdu, Deyang, Mianyang, Meishan, Leshan, Yibin, Zigong, Neijiang, Ziyang, Suining, and Nanchong) demonstrate significantly higher emergy investment and environmental loading ratios compared to mountainous regions (including Aba, Garzê, Liangshan, and Panzhihua), due to higher purchased emergy inputs (2.97 × 10²¹ sej/ha/year). These variations are key drivers of ecosystem differentiation, providing a theoretical foundation for optimizing regional agricultural production structures and reducing environmental loading (Lu et al., 2017; Liu et al., 2019). Based on the results of this study, we propose specific policy recommendations and technical improvement measures to help each region achieve efficient and sustainable agricultural production (Table 4).

The emergy environmental sustainability index provides a comprehensive measure of a system’s capacity for sustainable development, and regional variations in these indices reveal how different combinations of energy inputs affect system sustainability. Higher environmental sustainability index values indicate that a system has achieved favorable economic benefits and ecological balance while maintaining its core stability (Sha et al., 2015; Cheng et al., 2017). The sustainable development of Sichuan’s corn planting systems exhibits distinct spatial heterogeneity patterns, aligning with the ecosystem core hypothesis regarding the impact of purchased emergy inputs on ecosystem evolution (Wang and Zhai, 2019). Trend surface analysis revealed a decreasing gradient in the sustainable development index from northeast to southwest (R² = 0.63, p < 0.05), reflecting the combined effects of natural resource endowment and purchased emergy inputs. As noted, under identical climatic conditions, the coexistence of different ecosystem types depends on the relationship between natural and purchased emergy (Wang and Zhai, 2019). The higher sustainable development index in northeastern Sichuan likely results from a better balance between favorable natural conditions and moderately purchased emergy inputs. Conversely, the lower index in the southwest may indicate an overreliance on purchased emergy inputs in pursuit of high yields. In the northern farming-pastoral ecotone, cash crops required the highest purchased emergy input (101.04–147.67 × 10¹⁴ sej/ha/year), while natural grasslands were able to maintain functionality with minimal purchased emergy input (3.53–4.21 × 10¹⁴ sej/ha/year). Higher purchased emergy inputs may compromise long-term ecosystem sustainability (Wang and Zhai, 2019). By regulating purchased emergy inputs, we can guide the system’s evolution toward greater sustainability while maintaining the stability of the natural emergy core.

Sensitivity analysis results indicated that FN and N showed significant negative correlations with ESI, while FR and R demonstrated positive correlations with ESI (Figure 10). FN has the most significant impact on ESI, with a 20% increase leading to a 13.7% decrease in ESI, showing the strongest negative sensitivity. The sensitivity of variables affecting ESI ranked as follows: FN > R > N > FR. Changes in FN produced the maximum fluctuation in the sustainable development index, while the impact of FR was relatively moderate. Reducing dependence on FN and N while increasing the use of FR and R will significantly enhance sustainable development levels. In particular, controlling FN usage should become the priority strategy for improving ESI.

From 2000 to 2022, corn production in Sichuan increased from 5.474 million tons to 10.4622 million tons, a total increase of 91.13% (Figure 11A). The cultivation area expanded from 1.2355 million hectares to 1.855 million hectares, an increase of 50.15% (Figure 11B). Sichuan’s corn cultivation exhibited distinct developmental phases: Initial Growth Phase (2000–2007), with production increasing by 18.97% and area by 10.84%, demonstrating preliminary scale expansion; Rapid Development Phase (2008–2015), with production growing by 47.06% and area by 29.57%, showing significant intensification; and Stable Adjustment Phase (2016–2022), with marginal decreases in both production (1.12%) and area (0.59%), indicating structural optimization trends. The minor decline in both production and area post-2016 suggests a transition toward quality-oriented and efficiency-focused production, reflecting the impact of agricultural supply side structural reform policies. The multi-indicator emergy analysis framework provides comprehensive theoretical support for optimizing regional agricultural production (Amiri et al., 2019; Lewandowska-Czarnecka et al., 2019). By rationally regulating purchased emergy inputs while maintaining the stability of the natural emergy core, we can guide the system toward greater efficiency and sustainability (Patrizi et al., 2018).

This study has several limitations that require improvement. First, regarding the completeness of the indicator system, while 18 evaluation indicators were established, the study did not consider factors such as climate change and extreme weather events affecting corn planting systems. Additionally, the lack of long-term monitoring data for ecological elements, including soil quality evolution and groundwater level changes, may have led to underestimations of the actual environmental carrying capacity pressures. Regarding temporal scale limitations, owing to data availability constraints, the study lacks a long-term dynamic assessment. Although corn yield and area changes from to 2000–2022 were analyzed, emergy indicators for these years were not incorporated into the analytical framework, preventing a comprehensive understanding of sustainability evolution patterns in Sichuan’s corn planting systems. Regarding spatial scale limitations, the analysis using prefectural-level cities/autonomous prefectures as basic units overlooks county-level heterogeneity. Future research should focus on the system vulnerability at additional microscales. While this study identified significant regional heterogeneity in corn agroecosystem patterns, the underlying socio-economic and biophysical drivers remain poorly understood. The relative contributions of anthropogenic factors (agricultural policies, market dynamics and farmer decisions) versus environmental determinants (soil properties, climate variability, landscape structure) require systematic quantification. In the future, we should adopt integrated socio-ecological approaches to: (1) determine the relative importance of these driver categories, (2) identify threshold effects and interactions, and (3) develop predictive models for agroecosystem sustainability under varying conditions. Such mechanistic insights would advance both theoretical understanding and evidence-based management strategies for sustainable corn production.

Acknowledging these limitations, recent technological advances offer promising solutions for future research. Remote sensing data integrated with artificial intelligence models have demonstrated significant potential in agricultural monitoring and assessment, providing dynamic, high-resolution spatiotemporal information that can overcome the constraints of static statistical data (Omia et al., 2023; Correa et al., 2024; Neri et al., 2024; Song et al., 2024a; Song et al., 2024b). Machine learning algorithms applied to satellite imagery can accurately estimate crop yields, monitor real-time agricultural practices, and assess environmental impacts at fine spatial scales, thereby enabling more comprehensive and timely sustainability evaluations. These emerging technologies could complement emergy analysis by providing continuous monitoring capabilities and reducing reliance on annual statistical compilations.