Agricultural activities contribute substantially to global greenhouse gas (GHG) emissions, posing significant challenges to global food security and climate change mitigation. China, as the world's largest emitter of carbon dioxide, faces considerable pressure to reduce its carbon footprint, including emissions from its agricultural sector, which accounts for 17% of the total. While the industrial sector remains the primary contributor, reducing agricultural carbon emissions (ACEs) is a crucial national priority. The digital transformation of agriculture (DTA) presents a potential avenue for achieving this goal by optimizing resource utilization, improving efficiency, and promoting sustainable practices. This study addresses the critical question of whether and how DTA influences ACEs in China. The research aims to empirically analyze the carbon emission reduction effect of DTA, exploring the underlying mechanisms and accounting for regional heterogeneity. The study's findings hold significant academic and practical implications, informing policy decisions aimed at achieving agricultural low-carbon development in China.

Existing literature has explored the measurement of digitalization levels across various sectors and regions, including some studies focusing on China's inter-provincial and city-level digitalization. However, research specifically measuring digitalization in agriculture is less prevalent. Studies focusing on the digital economy's impact on carbon emissions generally suggest a positive role in reducing emissions, although some studies indicate a more complex, potentially non-linear relationship. Previous research has also identified several factors influencing agricultural carbon emissions, such as industrial structure, population size, economic development, energy intensity, and consumption patterns. While some studies have examined the impact of digital finance on ACES, a comprehensive investigation of the broader influence of DTA on ACEs, encompassing mediating mechanisms and accounting for regional heterogeneity, remains scarce.

This study employs a panel data analysis using data from 30 Chinese provinces (excluding Tibet, Hong Kong, Macau, and Taiwan) between 2015 and 2021. The study constructs a comprehensive index for DTA based on three dimensions: digital infrastructure, digitalization of the agricultural industry, and digital finance, using the entropy weight method. Agricultural carbon emissions (ACE) are calculated based on various carbon sources including fertilizers, pesticides, agricultural film, diesel oil, cultivation, and irrigation. The benchmark regression model examines the direct relationship between DTA and ACE, controlling for relevant factors such as urbanization, economic development, agricultural disasters, fiscal expenditure, and agricultural industrial agglomeration. A mediating effect model is used to explore the mechanisms through which DTA affects ACEs, considering the mediating roles of agricultural production scale, agricultural industrial structure optimization, and agricultural technological progress. A threshold effect model is then employed to investigate the potential nonlinear relationship between DTA and ACE, accounting for the heterogeneity across regions and industries. Robustness checks are conducted through instrumental variable estimation and different econometric model specifications. Finally, the study undertakes a heterogeneity analysis to identify the impact of DTA on ACEs across different regions (east, central, west, northeast), in major grain-producing areas versus non-grain producing areas, and for different sources of carbon emissions.

The key findings of the study are as follows: 1. **Significant Negative Impact of DTA on ACEs:** The benchmark regression consistently shows a significant negative relationship between DTA and ACEs, confirming the hypothesis that DTA reduces ACEs. The fixed-effects model proves superior to OLS and random-effects models. 2. **Mediating Mechanisms:** The mediating effect model reveals that DTA reduces ACEs through three pathways: agricultural production scale expansion, agricultural industrial structure optimization, and agricultural technological progress. Agricultural industrial structure optimization exhibits the strongest mediating effect, while scale expansion has the weakest effect. 3. **Heterogeneity:** The study finds significant regional heterogeneity in the impact of DTA on ACEs. DTA primarily reduces ACEs in the eastern region and non-grain production areas. For carbon emission sources, the effect is most prominent for fertilizers and diesel oil. 4. **Nonlinear Relationship and Threshold Effect:** A threshold effect model indicates a nonlinear relationship between DTA and ACEs. Below a certain threshold, DTA's impact on ACEs is insignificant; however, above this threshold, the inhibitory effect becomes significantly negative. In 2021, 24 provinces in China had DTA levels exceeding the threshold. 5. **Robustness:** Results remain robust after instrumental variable estimations and various model specifications. Relevant statistics and data points are presented throughout the study in tables and figures illustrating the levels of DTA, ACEs across different regions, and the results of regression models, mediating effect tests, and threshold effect analyses.

The findings of this study strongly support the hypothesis that DTA contributes significantly to reducing ACEs in China. The identification of three distinct mediating pathways highlights the multifaceted nature of DTA's influence on carbon emissions. The significant heterogeneity observed underscores the importance of considering regional contexts and industrial characteristics when implementing DTA policies. The nonlinear relationship and threshold effect suggest that a certain level of DTA is necessary to trigger substantial carbon emission reductions. This finding is particularly relevant for policymakers, highlighting the need for targeted interventions that focus on exceeding the identified threshold in regions where DTA levels are currently low. The results further highlight the importance of investments in digital infrastructure, agricultural technology, and human capital development to enhance DTA's impact on environmental sustainability.

This study demonstrates the significant role of DTA in reducing agricultural carbon emissions in China. The findings highlight the importance of considering mediating mechanisms, regional heterogeneity, and threshold effects when formulating policies to mitigate climate change in the agricultural sector. Future research could focus on refining DTA measurement, exploring additional mediating factors, conducting cross-country comparisons, and using finer-scale data (municipal or county-level) for more precise analysis. Further research exploring the long-term impacts of DTA on ACEs and the interaction effects of DTA with other climate change mitigation policies is also recommended.

The study's limitations include the availability of data, focusing on the provincial level, and the time frame of analysis (2015-2021). The chosen DTA index, while comprehensive, may not fully capture all aspects of digital transformation in agriculture. Furthermore, the study focuses primarily on China's context, limiting the generalizability of the findings to other countries with different agricultural systems and policy environments. Future studies could address these limitations by utilizing more granular data, exploring alternative DTA measurement approaches, and conducting cross-country comparisons.