Diversified cropping systems, particularly those incorporating legumes, are promoted for boosting food production while minimizing environmental impacts and input requirements. Legumes offer potential benefits through various ecosystem services, including pest and disease control, carbon sequestration, and soil fertility enhancement. Despite these advantages, legume cultivation has declined globally due to yield instability and lower yields compared to other crops, leading to reduced high-quality protein production and the loss of crucial ecosystem services. The type of legume used (grain, forage, or green manure) influences biomass production and residue quality, directly affecting the benefits for subsequent crops in the rotation. Therefore, effectively designing and implementing legume-based cropping systems holds significant potential for improving both local and global crop production while mitigating negative environmental effects. Previous research on the impact of legume pre-crops on subsequent crop yields has been limited in scope, focusing on specific crops or regions. This study addresses this gap by conducting a global-scale quantitative synthesis, analyzing a broad range of environmental conditions to determine the magnitude and variability of legume pre-crop effects. The study hypothesizes that nitrogen (N) fertilization for main crops will reduce the legume pre-crop effect due to suppressed nodulation and N2 fixation, and that crop diversification will decrease the non-N legume pre-crop effect. Understanding the key environmental and management factors influencing main crop yield after legume inclusion remains incomplete; thus, a comprehensive global-scale quantitative synthesis is crucial for informing better design and implementation of legume-based rotations.

Prior research on the effects of legume pre-crops on subsequent crop yields has been limited, often focusing on specific crop species or confined to regional scales. Studies have explored the impact of legumes on cereals, but a comprehensive global analysis incorporating diverse legume types and main crops has been lacking. The literature highlights the combined effects of nitrogen (N) provision and other non-N effects (e.g., pest suppression, improved soil properties) contributing to yield enhancement after legumes. Existing meta-analyses have either focused on particular crop species or restricted their scope to regional contexts. A global-scale analysis is therefore needed to address the growing interest in legume and crop diversification amongst researchers, farmers, and stakeholders in the agricultural value chain.

This study employed a systematic review and meta-analysis of data from peer-reviewed publications. The search encompassed Web of Science, Google Scholar, and CNKI, using keywords related to crop rotations, sequences, and yield. The search was conducted in October 2020 and followed PRISMA guidelines. Study selection criteria included field experiments with side-by-side comparisons of legume and non-legume pre-crop rotations, reported or calculable subsequent crop yield data, consistent initial conditions, and stated experiment location. Experiments were considered distinct if multiple experiments were reported in a publication; otherwise, different publications from the same experiment were coded as the same study. Data extraction included mean yields, number of replications, and indicators of precision (standard error or confidence intervals). Yields were standardized for all main crops in non-legume systems. The dataset covered 11,768 paired observations from 462 field experiments in 53 countries between 1959 and 2020. Site characteristics (location, climate, soil quality), management information (crop species, cultivation practices), and data from figures were extracted. The natural log of the response ratio (lnRR) was calculated to measure the effect size. Observations with zero yield values were excluded. A meta-regression model was fitted with 21 potential modifiers as predictors using the R package *metaforest*, a machine-learning algorithm robust to overfitting. Variable importance was determined using the Gini impurity index, and model parameters were optimized using 10-fold cross-validation. The legume pre-crop effect was modeled using a linear mixed-effects model, accounting for the non-independence of studies. Publication bias was assessed using Fail-Safe N analysis and funnel plot inspection. Results were back-transformed to express the percentage change in yield. Analyses were performed using R 4.0.3.

The meta-analysis revealed a significant positive effect of legume pre-crops on main crop yield, with an average global increase of 20.4% (median effect 10.2%; 95% confidence intervals 17.7-23.1%). This benefit was less than a previously reported estimate of 29% for grain legume pre-crop effects on cereals, but the current study includes a wider range of legumes and main crops. The positive effect was observed in 91% of cereal-based systems. A meta-forest analysis indicated that nitrogen (N) fertilizer rate and crop diversity were the most important moderators of the legume pre-crop effect. The yield advantage decreased by 7% for each 50 kg N ha-1 added, and increasing crop diversity by one unit led to a 3.5% decrease in yield advantage. This negative correlation with N fertilizer rate was consistent across various main crops. The yield benefit from legume inclusion was more pronounced in systems with lower crop diversity, suggesting that the benefits of legumes are more significant in less diversified systems. Greater legume pre-crop effects were observed in low-yielding environments and with low N inputs. The yield advantage declined sharply with increasing initial main crop yield, becoming negligible above the average yield. The increase in yield was considerably higher in Africa (43%) than in other regions like North America (19%) and Asia (12%). Specific cropping sequences also affected the yield advantage. The legume-cereal sequence showed a significant yield advantage of 21%, and legume species showed variable effects, such as pigeon pea (32.4%) exhibiting a much stronger positive effect than common bean (14.5%). Management practices like legume purpose, tillage, and soil parameters also influenced the legume pre-crop effect.

The findings confirm the significant yield advantage of legume-based rotations globally. The negative correlation between yield benefits and N fertilization highlights the potential for reducing N inputs in intensive cropping systems by optimizing legume pre-crop effects. The negative relationship between crop diversity and legume pre-crop effects indicates the greater importance of legumes in low-diversity systems. The enhanced effects in low-yielding environments and low N input scenarios highlight the critical role of legumes in improving yields in resource-constrained settings, such as Africa and organic agriculture. These results support the idea that reduced legume use is associated with stagnating wheat yields in Europe. The variable effects across legume and main crop species and management practices suggest that optimizing cropping sequences and management strategies could further maximize the benefits of legume-based rotations. While legume inclusion might reduce the area allocated to main crops, this negative effect can be offset by the significant yield increases observed.

This global meta-analysis demonstrates the substantial yield advantages of legume-based rotations, particularly in low-input and low-diversity systems. Nitrogen fertilization and crop diversity are key moderators of these benefits. Optimizing legume use, N fertilization, and crop diversity offers potential for enhancing global crop production while advancing sustainable agriculture. Future research should focus on optimizing cropping sequences and management practices to maximize legume benefits in various contexts.

The study relied on data from published research, which might be subject to publication bias. Data availability varied across studies, influencing the level of detail and analysis in some instances. The analysis primarily focused on yield as a measure of success, neglecting other potentially relevant outcomes like environmental impacts, economic benefits, and social considerations.