Global climate change necessitates a focus on agricultural carbon emissions, which constitute a significant portion (16-17%) in China. Traditional agricultural practices have led to environmental damage and threaten human health. China's agricultural ecological transformation is crucial for sustainable development and rural revitalization. The digital economy, with its scale of 50.2 trillion yuan in 2022 (41.5% of GDP), offers potential for improving resource allocation and reducing production costs in agriculture. Integrating the digital economy with agriculture can alleviate information asymmetry, optimize resource allocation, and reduce transaction costs. This study addresses three key questions: Does the digital economy promote agricultural ecological transformation? What is the mechanism of low-carbon technological innovation in this process? And, how does the effect of the digital economy vary at different levels of low-carbon technological innovation? The study aims to contribute by constructing an evaluation index system for agricultural ecological transformation, incorporating the digital economy as a driver, and considering the threshold effect of low-carbon technological innovation. This research utilizes a nonlinear dynamic panel threshold model to examine the heterogeneous effects across regions, offering new insights for low-carbon agriculture development.

Existing literature consistently highlights the digital economy's role in improving agricultural economic benefits, balancing environmental and economic benefits, and empowering agricultural ecological transformation through technological spillover effects. The digital economy enhances resource utilization, promotes technological innovation, and optimizes agricultural industry structure. However, research also notes the importance of low-carbon technological innovation, which is crucial for sustainable agriculture and green development in developing countries. While some studies explore the impact of low-carbon technological innovation on agricultural ecological transformation from an enterprise perspective or for specific sectors (forestry), a comprehensive analysis of its interaction with the digital economy in the context of China's agricultural sector is lacking. This study bridges this gap by focusing on the agricultural sector in China and using an improved dynamic threshold regression method to account for endogeneity and dynamic changes in the data, providing a more robust analysis than previous studies.

This study constructs an evaluation index system for agricultural ecological transformation using the Super-SBM model. This model addresses the issue of slackness in input-output variables and the problem of unexpected outputs. Input factors include labor, land, irrigation, fertilizers, pesticides, agricultural film, energy, and machinery. Expected output is agricultural output value, while unexpected output is agricultural non-point source pollution. The Super-SBM model is used to calculate the level of agricultural ecological transformation in 30 Chinese provinces from 2013-2021. A dynamic panel threshold model is then employed to analyze the impact of the digital economy (DE) on agricultural ecological transformation (AEE), with low-carbon technological innovation (LCI) as the threshold variable. Control variables include disaster rate (DR), urbanization rate (URB), agricultural internal structure (AIS), and rural human capital (HC). The digital economy index is calculated using the entropy method based on indicators related to digital infrastructure, digital industrialization, and industry digitalization. Low-carbon technological innovation is measured using the number of green patent applications. The dynamic panel threshold model considers the endogeneity and dynamic changes using system GMM dynamic methods.

The analysis reveals a low overall level of agricultural ecological transformation in China (average 0.607), although exhibiting a steady upward trend (average annual growth rate of 6.99% from 2013 to 2021). Significant regional disparities exist, with Guangdong Province showing the highest level (0.866) and Jilin Province the lowest (0.320). The dynamic panel threshold model indicates a significant double threshold effect of low-carbon technological innovation (LCI) on the relationship between the digital economy (DE) and agricultural ecological transformation (AEE). Below the lower threshold (LCI ≤ 4.283), the digital economy negatively impacts AEE. Between the thresholds (4.283 < LCI ≤ 4.642), a positive effect emerges, and above the higher threshold (LCI > 4.642), a significant positive impact is observed, showcasing a U-shaped relationship. Urbanization rate and disaster rate positively affect AEE, while rural human capital shows a negative (though insignificant) impact. The agricultural internal structure has a positive but insignificant impact.

The findings suggest a nuanced relationship between the digital economy and agricultural ecological transformation, contingent on the level of low-carbon technological innovation. At low levels of LCI, the digital economy may not be effectively utilized due to technological limitations and mismatches between technology development and digital economy implementation. However, as LCI surpasses a critical threshold, the digital economy's positive impact is amplified through efficient resource allocation, knowledge spillover, and technological advancements. This highlights the importance of investing in LCI to unlock the full potential of the digital economy in promoting agricultural ecological transformation. The regional variations underscore the need for targeted policies tailored to different levels of technological development.

This study demonstrates a U-shaped relationship between the digital economy and agricultural ecological transformation in China, contingent upon low-carbon technological innovation. Policy recommendations include strengthening digital infrastructure, fostering digital technology innovation and talent cultivation, promoting regional coordination in digital economic development, and focusing on improving low-carbon technological innovation levels, especially in regions lagging behind. Future research could expand the data dimensions to include enterprise or micro-level analysis and explore the influence of other regional heterogeneity factors, such as environmental regulations and industrial integration.

The study utilizes regional macro-level data, limiting the analysis to provincial levels. Future research could benefit from incorporating micro-level data or finer geographical units. Additionally, the study focuses primarily on low-carbon technological innovation as a threshold variable, while other factors such as environmental regulations or industrial structure might also play a significant role. Further research could explore these factors and their interplay with the digital economy in driving agricultural ecological transformation.