Governing non-grain production of farmland: A differentiated strategy based on grain production loss risk

Against the dual backdrop of escalating global grain security risks and deepening urban-rural development transformation, coordinating grain security with increased farmer income has become a critical challenge in land management. The non-grain production of farmland (NGPF) phenomenon epitomizes this contradiction. Current governance approaches primarily focus on managing existing issues and implementing universal policies, lacking proactive prediction and forward-looking warning for potential occurrence zones and the resulting grain security risks. Therefore, this study developed an integrated analytical framework comprising “probability prediction—risk assessment—categorical governance.” Taking Hubei Province as a case study, the research area was divided into three zones based on agricultural functional differences: the urban agriculture zone (UAZ), the major grain-producing zone (MGZ), and the characteristic agricultural products zone (CAPZ), with research conducted separately for each zone. Specifically, this study first predicted the NGPF probability using the Maxent model and analyzed its driving mechanisms. Subsequently, it combined the probability-loss assessment model to quantify the potential risks of NGPF to unit and total grain yields. Finally, for medium-to-high-risk areas, the SOM+k-means clustering method was employed to identify dominant driver bundles and formulate differentiated governance strategies accordingly. The results indicate that NGPF exhibits significant regional heterogeneity: UAZ is primarily driven by irrigation convenience (16.4 %), mechanization level (16.5 %), and aggregation index (14.6 %), exhibiting the highest probability of NGPF conversion; CAPZ is primarily driven by the combined effects of distance from town (13.6 %), proportion of aging agricultural labor force (17.6 %), and percentage of people with contracted management rights (10 %), ranking second in probability; MGZ is jointly driven by mechanization level (27.1 %), percentage of permanent basic farmland (17.6 %), proportion of aging agricultural labor force (12.3 %), and average educational attainment of the rural population (11.7 %), exhibiting the lowest probability. Risk analysis reveals that unit grain yield loss risk follows the pattern of UAZ > MGZ > CAPZ. MGZ, characterized by large farmland sizes and a high multiple crop index, exhibits the most pronounced total grain yield loss risk, with a 30 m raster scale mean of 115.59 kg, followed by UAZ (103.82 kg) and CAPZ (68.55 kg). Based on these findings, the study further subdivided medium-to-high-risk farmland across the three agricultural zones into spatial governance units with clearly dominant risk mechanisms, proposing targeted governance measures. This framework provides support for the forward-looking and differentiated governance of NGPF.