The study was conducted in Tianshui Village, Quzhou County in the North China Plain ( Supplementary Figure S1 ). This region has a temperate semi-humid continental monsoon climate, with an annual average temperature of 13.1°C and a frost-free period of 201 days. The average annual precipitation is 542.8 mm, with a distinct wet season from late June to late September, during which 66.7% of the annual precipitation occurs. The warm climates support double cropping within a single year, with the predominant cropping system being winter wheat-summer maize rotation, where the wheat is generally sown in October and harvested in June of the following year. The soil texture comprises light loam, medium loam, sandy loam, clay and salt-affected. The basic properties of the surface layer soil (0–20 cm) are as follows: total nitrogen 940 mg kg⁻¹ (from700 to 1060 mg kg⁻¹), available phosphorus 23.0 mg kg⁻¹ (Olsen-P, from 1.4 to 50.4 mg kg⁻¹), and exchangeable potassium 132 mg kg⁻¹ (Exc-K, from 67.1 to 190 mg kg⁻¹) ( Supplementary Table S1 ).

The CLUP was established by the Science and Technology Backyard (STB) – a platform established by China Agricultural University to foster information and knowledge exchange between the scientific institutions and farming communities. This project aims to collaborate smallholder farmers, local government, and agricultural enterprises, with the final goals of consolidating fragmented farmlands and implementing tailored practice packages ( Figure 1 ). The CLUP implementation process involves five key steps:

Step 1: Determine the location. We selected a large-scale farmland consisting 23 adjacent plots from different smallholder farmers, covering a total area of 2.7 ha. We accordingly established CLUP on this farmland.

Step 2: Select the leading farmers. Farmers with extensive experience in cultivating farmlands and organizing activities were selected as leading farmers. They participated in CLUP activities either voluntarily or by invitation. Leading farmers played a crucial role in assisting CLUP staff to in adapting recommended management practices to local conditions and were responsible for overseeing field production.

Step 3: Consolidate farmlands. Government and leading farmers encouraged other smallholder farmers through meetings, broadcasts and home visits. Participation in CLUP was voluntary, with a commitment not to alter their land property rights.

Step 4: Develop tailored practice packages. Scientists and participating smallholder farmers got together to discuss and revise optimal practice proposed by scientists. The meetings, generally held by scientists, began with an introduction of the suggested practices, including their operating specifications, effects, and risks. This was followed by collecting feedbacks from smallholder farmers based on their practical experience and concerns. Finally, the tailored practice packages were determined through a bidirectional interaction and compromise between scientists and farmers.

Step 5: Implement tailored practice packages. The practice packages were jointly implemented by the multi-actors, including government, universities, enterprises, and smallholder farmers. Under government coordination, fragmented farmlands owned by smallholder farmers were consolidated into large-scale farmlands for demonstrating CLUP. There were two ways to implement tailored practice packages – unified operation and personal operation. For unified operation, leading farmers centrally managed activities such as land preparation, sowing, first fertilization, and plant protection. They coordinated the unified purchase of products, equipment, and services from enterprises. Consequently, economic costs were substantially reduced through bulk purchasing of seeds and fertilizers. Personal operation was generally applicable to labor-intensive activities, such as topdressing and irrigation, which were carried out by individual farmers. During crop production process, smallholder farmers received technical training and field guidance provided by STB staff. At the harvest time, all the CLUP activities were temporarily halted, allowing each individual to harvest independently and retain their own grains to avoid economic disputes that could affect the subsequent works. It is worth mentioning that throughout the entire production process, all participating farmers have the right to supervise the operations of CLUP.

To compare the effectiveness of CLUP, we set other two treatments – scientist practices (SP) and conventional farmer practices (FP) over the three consecutive years. SP adopted the best management practices (i.e., integrated soil-crop system management, ISSM) to achieve high-yield and high-efficiency in Quzhou Experimental Station (Chen et al., 2011). Plot size of SP was 1800 m² with four replications. Simultaneously, we selected 20 adjacent farmlands (average 0.2 ha per household) with same soil conditions and similar sizes as those in CLUP for the FP treatment. The agricultural operations in FP treatment were carried out entirely according to farmers’ conventional practices. Both CLUP and FP conducted monitoring and sample collection in each of the 23 and 20 groups of participating farmers’ plots. Detailed management information is elaborated in Table 1 .

Significant variations were observed in the N flow of wheat production among the three groups ( Figure 2 ). The total N input in FP, which included deposition, irrigation, chemical fertilizer, seed and biological fixation, reached 298 kg ha^-1, with 84% derived from chemical fertilizers. In comparison, N input in the CLUP and SP was reduced by 19% and 31% respectively, primarily due to notable reductions in fertilizer inputs (22% and 35%). Additionally, SP effectively utilized up to 15 kg ha^-1 of residual nitrogen in the soil, thus mitigating the risk of nitrogen surplus. While wheat nitrogen uptake in CLUP (188 kg ha^-1) was slightly lower than SP (194 kg ha^-1), it significantly exceeded the FP by 11%. Notably, Nr loss (comprising N leaching, N₂O emissions, NH₃ volatilization, and N accumulation) was substantially lower in CLUP and SP (53 kg ha^-1 and 12 kg ha^-1, respectively) compared to FP (127 kg ha^-1).

Higher grain yield and N recovery efficiency were observed in CLUP and SP ( Figure 3 ). Specifically, grain yield in CLUP and SP were 8 t ha^-1 and 8 t ha^-1, marking increases of 13% and 17% respectively compared to FP. N recovery efficiency was 58% in FP, while CLUP and SP achieved 78% and 95%, respectively. However, there still exists a 4% and 18% gap in yield and N recovery efficiency between CLUP and SP.

The average GHG emissions per unit area of CLUP were 4497 CO₂ ha^-1, which fell between those of FP and SP ( Figure 4a ). The main sources of GHGs were nitrogen fertilizer production and electricity used for irrigation, contributing 34% to 41% and 22% to 34% of the total emissions across the three production models, respectively. Phosphate fertilizer application contributed an average of 16% of GHG emissions, followed by N fertilizer and K fertilizer, fuel, production and transportation of herbicide. Notably, GHG emissions from fuel consumption in CLUP were reduced by 19% compared to SP and by 14% compared to FP. FP had the highest GHG emissions per unit yield (827 kg CO₂ eq-Mg ^-1) among the three production models, with significant reductions of 32% and 52% for CLUP and SP, respectively ( Figure 4b ). The NECB ranged from 3 to 40 Mg C ha^-1 across the three groups ( Figure 4c ), with CLUP showing a 20% increase compared to FP, mainly due to the significant increase in gross primary productivity.

Similarly, the water eutrophication potential of CLUP was positioned between that of FP and SP ( Figure 5a ). CLUP’s water eutrophication potential (17 kg PO₄-eq ha^-1) decreased by 21% compared to FP (22 kg PO₄-eq ha^-1), but still showed a 15% difference compared to SP (15 kg PO₄-eq ha^-1). Nitrogen fertilizer application contributed 72% to 77% of PO₄-eq across all three production models and was the main cause of water body eutrophication, followed by fuel consumption, which accounted for 13% to 21%.

The soil acidification potential followed a similar trend to water eutrophication ( Figure 5b ). CLUP’s soil acidification potential (74 kg SO₂-eq ha^-1) was 188% lower than that of FP. In three production models, Nitrogen fertilizer application emerged as the main contributor to soil acidification, accounting for 47% to 51%, followed by electricity used for irrigation. Notably, SP’s secondary source of soil acidification shifted to fuel, due to substantially reduced irrigation amounts, which constituted 39% of the total.

CLUP achieved a commendable high net ecosystem economic budget of 6,986 CNY ha^-1 with minimal cost and labor inputs, comparable to SP and FP ( Figure 6a ). This success was mainly due to high wheat yields and low environmental damage costs, including cumulative costs from greenhouse gas emissions, water eutrophication, and soil acidification.

Farming costs accounted for 84% to 89% of the total production cost across all three models. In the CLUP model, farming costs were 8% lower than FP and 9% lower than SP, mainly due to reduced fertilizer purchase prices ( Figure 6b ). Detailed cost expenditures are provided in Supplementary Table S3 . Additionally, environmental damage costs in the CLUP model were 1,477 CNY ha^-1, 28% lower than FP but 17% higher than SP. Remarkably, CLUP exhibited outstanding agricultural labor productivity, increasing by 53% compared to FP and 60% compared to SP ( Figure 6c ). This highlighted the efficiency gains achieved through CLUP adoption, streamlining operations, minimizing labor inputs, and maximizing productivity.

Smallholder farmers’ adoption of practices is influenced by various factors, involving challenges related to accessibility, usability, and understandability of technologies (Eastwood et al., 2020; George, 2014; Wens et al., 2021). Consequently, these challenges often result in low efficacy of technologies in smallholder field plots (Ayisi et al., 2022; Zhao PF et al., 2016). Engaging smallholder farmers in the process of innovating practices, combining scientific knowledge with smallholder farmers experience to develop tailored technologies, has proven effective in fostering farmers’ willingness to adopt new practices (Colnago et al., 2021; Fieldsend et al., 2022; Maryono et al., 2024). In this study, leading farmers, acting as farmer representatives, were invited from the outset to participate in working sessions aimed at developing practice packages. This approach provided them with a clear vision and understanding of the various management programs being developed (Wellens et al., 2013). Concurrently, scientists gained insights into the field realities that needed integration into the programs. The practice was iteratively refined through dialogue between scientists and farmers, striking a balance between their respective needs. For instance, the nitrogen fertilizer amount was adjusted from the recommended 162 kg N ha^-1 to 196 kg N ha^-1. This modification, although suboptimal, stayed below the FP average of 250 kg N ha^-1 and addressed farmers’ concerns about potential yield penalties from reduced chemical nitrogen use ( Table 1 ). Through this typical participatory research, smallholder farmers were able to access, use and understand the practice. Field application results showed that the revised practice system resulted in a 13% increase in grain yield, a 35% improvement in nitrogen recovery efficiency, and a 18% reduction in GHG emissions compared to FP. This evidence underscores that significant potential of the model to enhance the sustainability of crop production for predominantly smallholder farmers.

Implementing advanced agronomic practices on the fragmented lands is crucial for achieving sustainable production among smallholder farmers (Duan et al., 2021; Zhang et al., 2022b). Agricultural intensification has been identified as a potential solution (Maryono et al., 2024; Sánchez et al., 2018). Most studies have examined the models and effectiveness of large-scale operations through land transfer. For example, Zhang et al. (2022b) combined public-private partnerships with large-scale farming, a model that resulted in a 12% increase in yield and a 33% reduction in carbon footprint. Similarly, Zhang et al. (2022a) proposed a comprehensive straw management strategy for smallholders, which led to a 10% increase in yield. However, these solutions were difficult to replicate due to the low willingness of smallholder farmers to transfer land, particularly in developing countries (Wang et al., 2023; Xu et al., 2023). Therefore, innovating large-scale land management while safeguarding the land rights and interests of smallholder farmers is crucial to facilitate the adoption of advanced practices.

This study divided field operations into unified practices and personal practices, considering the characteristics of each practice and the specific needs of farmers. Unified practices involved overcoming land constraints and centralizing agronomic practices at scale across all farmer plots, coordinated by scientists and leading farmers. This approach aimed to ensure the application of mechanized ploughing, sowing, and drone spraying in fragmented plots, and alleviate economic costs associated with their application. Practices that did not require large-scale farm machinery, such as irrigation and top-dressing, were conducted by farmers themselves according to the recommended practice packages. During harvesting, farmers independently harvested their plots to protect their interests. This model facilitated large-scale operations without necessitating land transfer, thereby preserving land property rights. Overall, the model effectively integrates farmers’ needs, technological advancements, and land characteristics. It combines theoretical knowledge and operational practice packages, facilitating the integration of fragmented land ownership and the application of large-scale practice. As such, it serves as a convincing and successful example of achieving sustainable production despite the restrictions of land transfer.

Establishing transformative partnerships is crucial for successfully implementing advanced practices on a large scale across fragmented farmlands. This involves not only enhancing farmers’ willingness to adopt practices but also establishing a comprehensive support system. Such support system includes policy guarantees for land rights, access to advanced practices, and provision of large machinery and equipment (Adnan et al., 2019). However, these requirements often exceed the capacity and resources of individual smallholder farmers (George, 2014).

In the CLUP, key stakeholders underwent significant role transformations to enhance collaboration and ensure the project’s success. Traditionally, government-led initiatives have been characterized by top-down approaches, often leading to inefficiencies and failing to address the real needs of farmers (Zhang et al., 2018). However, in this study, the government transitioned from a dominant regulatory role to a supportive one, focusing on securing land tenure and introducing quality control measures, such as labeling on agricultural inputs to protect farmers from substandard products. Similarly, private enterprises shifted from supplying generic agricultural inputs to offering tailored products and services that better met farmers’ specific needs. This transformation not only promoted economic development within the private sector but also built trust between service providers and farmers. Moreover, scientists evolved from merely offering technical recommendations to becoming active collaborators, working closely with farmers to solve field-specific challenges. Moreover, farmers involved in CLUP took on supervisory and organizational roles within the project, mitigating the potential for profit-driven opportunism in the private sector (Yin et al., 2024). These changes in stakeholder roles created a more effective and mutually beneficial multi-actor partnership, driving both productivity and sustainability in smallholder agriculture.

Under the collaborative efforts of multi-actor partnership, advanced agricultural technologies have been implemented among smallholder farmers. Optimizing nitrogen fertilizer application played a key role in enhancing both crop yield and environmental sustainability. By adjusting the timing and amount of nitrogen applied, the nitrogen supply was better synchronized with the crop’s growth stages. This ensured that nutrients were available during critical periods, when crops could most efficiently utilize them, while minimizing losses through leaching and volatilization (Tian et al., 2024). In addition to fertilizer optimization, changes in tillage practices further contributed to both increased productivity and reduced environmental impact. The shift from traditional rotary tillage to a combination of deep tillage followed by rotary tillage enhanced soil structure by improving water infiltration and root growth. This not only allowed for more efficient nutrient uptake but also increased soil moisture retention, reducing the need for excessive irrigation (Feng et al., 2020). As a result, water use was optimized, leading to better resource management and a reduction in energy use for irrigation. The adoption of mechanized farming practices, such as large-scale sowing machinery and drone-based pesticide spraying, provided another layer of improvement. Mechanized sowing increased planting efficiency, allowing farmers to meet the recommended sowing window, even with limited labor availability. This helped align sowing with local climatic conditions, resulting in better crop establishment and higher yields. Additionally, drone-based pesticide spraying reduced labor inputs and ensured precise application, thereby minimizing chemical runoff and the overuse of pesticides, which further decreased environmental pollution (Arakawa and Kamio, 2023).

Ultimately, multi-objective crop production, grain yield, N recovery efficiency, net ecosystem economic benefit and agricultural labor productivity were improved by 14%, 35%, 86%, and 53%, respectively, and GHG emissions per unit area and per unit yield were reduced by 18% and 32%, respectively, highlighting the effectiveness of tailored agronomic practices facilitated by multi-actor participation ( Figure 7 ). The transformative partnership forged through CLUP integrates cutting-edge crop nutrition knowledge and policy backing from the public sector with efficient agricultural machinery services provided by the private sector. This approach transcends mere collaboration, it represents a comprehensive strategy for sustainable intensification in crop production dominated by smallholder farmers, representing a significant stride towards agricultural sustainability.

Globally, 500 million smallholder farmers contribute 28% of the food supply at a huge environmental cost (Bizikova et al., 2020; Liu, 2023). Enhancing their yields, incomes, and environmental performance is crucial for ensuring global food security and sustainable development (Yin et al., 2024; Zhang et al., 2016). Beyond China, other rapidly developing economies such as India, Brazil, and Indonesia are also facing similar conditions and challenges to transform smallholder agriculture towards sustainability (Mueller et al., 2013; Ricciardi et al., 2018). Multi-actor collaboration has gained increasing attention, such as Farmer Field Schools in Asia and Africa, the “LIAISON” project in Europe, and Soil Health Cards in India (Smyth et al., 2021; Fieldsend et al., 2022; Reddy, 2019). This study demonstrates a compelling example of multi-actor collaboration in achieving large-scale application of optimal management practices, thereby promoting sustainable intensification of agriculture. Consequently, CLUP has great potential for widespread implementation across various countries and regions.

Besides the traditional “top-down” strategy (e.g., policy incentives and financial support), a “bottom-up” approach is more crucial for implementing CLUP (Ensor and de Bruin, 2022). This approach relies on educating and raising awareness among farmers themselves. The STB organized training programs and workshops to disseminate information on best management practices (e.g., tillage and fertilization methods) and encourage farmers to participate in CLUP. Addressing actual needs and interests of farmers is essential for adopting innovative technologies. Guiding farmers to act as co-designers of technological innovations, rather than passive recipients, is vital for the rapid adoption of advanced technologies (Cui et al., 2018; Zhang et al., 2016; Mariano et al., 2012). Moreover, leading farmers play a vital role in disseminating optimal practices, serving as a conduit for advanced technologies, and connecting farmers with research institutions.

This study has two major limitations. First, the scale of its application is relatively small, because it was designed primarily as an exploratory experiment to demonstrate the feasibility and effectiveness of CLUP. In future studies, we will apply the CLUP model in other smallholder-dominated areas to verify its replicability and provide empirical evidence for subsequent upscaling. Additionally, this study focused primarily on the environmental impacts of nitrogen fertilizer application. Future research should encompass a broader and more comprehensive range of environmental impacts to provide a more accurate assessment of the CLUP model’s effectiveness.

Thirdly, despite our efforts to develop tailored agronomic practices and build a transformative partnership for the application of tailored practices, gaps still existed between CLUP and SP in terms of multi-objective coordination. It is critical to recognize that this process requires multiple rounds of improvement to continuously bridge the gap between CLUP and SP. Additionally, more innovations (e.g., collaboration methods, consolidation methods of fragmented farmlands, and involvement of more stakeholders) are needed to develop various flexible solutions for other regions worldwide with diverse endowment and situations.