Global spatially explicit yield gap time trends reveal regions at risk of future crop yield stagnation

The Green Revolution dramatically increased global crop production from the late 1960s to 2000, but this came at an environmental cost. Meeting increasing food demands while minimizing environmental impacts is crucial. Closing yield gaps—the difference between actual and attainable yields—is a key strategy for increasing production sustainably. Previous yield gap assessments often focused on single years, neglecting temporal variations. This study aims to provide a spatially and temporally comprehensive analysis of yield gaps, identifying regions at risk of future crop yield stagnation and informing policies to enhance food security and sustainability. The study focuses on ten major crops, accounting for 83% of global calories, analyzing data from 1975 to 2010 at a high spatial resolution. The researchers define 'attainable yield' as the 95th percentile of observed regional yields, representing the highest yield achieved under specific biophysical conditions. This approach builds upon and expands previous yield gap analyses, which have either lacked temporal depth or global coverage, thus limiting their policy implications. The study will develop a typology to categorize regions based on their yield gap trajectories and examine the correlation between yield gap closure and subsequent stagnation.

Existing literature on yield gaps mostly presents snapshots in time, hindering policy recommendations. Studies are categorized as 'narrow scope, high detail' (analyzing specific interventions) or 'broad scope, low detail' (addressing sustainable production possibilities). While many studies assess sustainable production based on static yield gaps, few translate these into policy recommendations. Some studies have examined yield gap trends over time, but these lacked consistent global coverage or focused on limited crops/regions. This study aims to overcome these limitations by providing a comprehensive analysis across multiple crops, high spatial resolution, and global coverage to draw policy-relevant conclusions on global production potential and needed interventions for food security.

The study calculates spatially explicit global time trends in attainable yields and yield gaps for ten major crops (maize, wheat, rice, oil palm, soybean, barley, sugarcane, sorghum, rapeseed, and cassava) from 1975 to 2010. A quantile regression model with year-specific coefficients is used to estimate area-weighted 95th percentile yields, representing attainable yields based on climate, soil characteristics, and irrigation. This method extends climate analogue approaches by incorporating a broader set of biophysical variables and producing continuous yield surfaces. The analysis uses a high-resolution historical crop dataset derived from census and survey information across approximately 20,000 political units. A static climatology is used to improve accuracy in modeling yield gap trends. Growth is quantified as the percentage of linear change relative to 2000 yield values. The study reports results globally and for eight geographical regions, with detailed country-level results for major producers in the Supplementary Information. The methodology addresses several limitations of previous studies by employing a larger dataset, finer spatial resolution, and a comprehensive temporal framework. The choice of a static climatology is justified by its superior performance in temporal cross-validation compared to using annual climate data. Key variables included in the model are growing degree days, mean annual precipitation, precipitation concentration index, irrigation fraction, available water capacity, soil organic carbon, pH, slope, and a vernalization factor for wheat. Model selection is performed using a stepwise approach and an iterative cross-validation procedure to avoid spurious time trends. The study utilizes a bootstrap method with 1,000 random samples to determine confidence intervals for model predictions. Time to yield gap closure is calculated based on linear regression of annual yield gaps over the interval 1998-2012, and stagnation probabilities are determined using piecewise linear regressions.

Attainable yields increased from 1975 to 2010 for most areas and crops, with maize, rapeseed, and soybean showing consistently high growth rates. Yield gaps increased for many crops and areas, particularly for maize and soybean. Globally averaged yield gaps increased for several crops, including barley, maize, rapeseed, rice, soybean, and wheat, but remained unchanged for others. The study introduced a three-category typology of yield gap trajectories: 'steady growth' (both actual and attainable yields increase), 'stalled floor' (attainable yield grows, actual yield stagnates), and 'ceiling pressure' (yield gaps close and/or attainable yield stagnates). Over 60% of maize area showed 'steady growth', in contrast to only 12% for rice. Rice and wheat showed high percentages of 'ceiling pressure' (84% and 56%, respectively). Notably, 'ceiling pressure' was strongly correlated with subsequent yield stagnation. The study revealed that areas with closing yield gaps (particularly rice in Asia, wheat in Europe, and sugarcane in Thailand) are at increased risk of future yield stagnation. The findings highlight significant regional heterogeneity in yield gap changes, with some regions showing decreases in yield gaps while others show increases. Maize and soybean, which provide a significant portion of calories indirectly or directly, show significantly increasing yield gaps unlike wheat and rice, which deliver a greater proportion of calories as food, showing no significant change in yield gaps. This discrepancy is potentially linked to higher investment in maize and soybean. The study also found that yield ceilings have decreased in most crops compared with a fixed-area counterfactual, implying a shift in production to less productive areas. The analysis suggests that the continued growth of attainable yields will require sustained investment in agricultural technologies.

The findings challenge the debate on the potential for future yield growth. While both actual and attainable yields have improved over decades, the nature of this growth varies across crops. The results suggest that rice is nearing a yield ceiling, necessitating technological innovation. Conversely, maize trends show ongoing improvement, implying continued potential for economic incentive-driven growth. The study’s typology of yield gap trajectories—steady growth, stalled floor, and ceiling pressure—provides a nuanced understanding of the drivers of yield changes. 'Ceiling pressure' regions require investment in technologies to raise attainable yields, while 'stalled floor' regions need focus on adoption of best practices. The study’s identification of regions at risk of yield stagnation highlights the need for targeted interventions to enhance food security. The method used in this study helps predict the likelihood of yield stagnation more reliably than previous methods based on local yield time series or yield gap size. This research emphasizes the policy relevance of identifying yield gaps and categorizing their evolution, although it doesn't offer specific prescriptions on how to address these gaps. The study also notes that focusing solely on staple crop yields could negatively impact nutritional diversity and food security.

This study provides a comprehensive, spatially and temporally explicit analysis of global yield gaps, revealing a strong correlation between yield gap closure and subsequent stagnation. The developed typology helps identify regions at risk and guide targeted interventions. The findings highlight the need for continued investment in agricultural research and development, particularly for crops crucial to food security. Future research could refine the model by incorporating additional factors (e.g., pathogen risk, adaptation to climate change) and explore the socio-economic factors influencing yield gap trajectories.

The study's attainable yield estimates may overestimate yield gaps in regions with limited data or specific limitations (e.g., soil rooting depth in Africa). The model's reliance on historical data may not fully capture the impact of future climate change or technological breakthroughs. The study also notes that the concept of closing yield gaps is not value-neutral and the approach doesn't offer prescriptions on how to address those gaps, focusing instead on identifying them and categorizing their evolution. While the analysis has global scope, local-level factors may influence yield gaps beyond the scope of this study.