Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 license. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. 1. Regenerative agriculture (RA) and the ‘Win-Win’ narrative Reducing the environmental impact of agriculture is central to climate and biodiversity policy (Ortiz et al 2021). Against this backdrop, RA has emerged as a proposed solution, promoted to improve soil health, reduce external inputs, and deliver climate benefits while maintaining productivity (Sher et al 2024). The concept features widely in corporate sustainability strategies, voluntary standards, and policy discussions (e.g. McDonald’s 2024), and is often presented as a ‘win-win’ approach for food production and the environment (Wilson et al 2024). However, RA is not a codified management system. Unlike certified organic, it lacks a consistent definition and has been described as a philosophy, a set of principles, and a collection of management practices (British Ecological Society 2025, Congreves 2025). Commonly cited principles include minimising soil disturbance, maintaining soil cover, maximising plant diversity, sustaining living roots, and integrating livestock (Kipling et al 2025). However, practices vary considerably, making generalisations about impacts difficult. Reported yield effects range from negative to positive (Krauss et al 2020, Li et al 2023), while greenhouse gas (GHG) impacts vary depending on changes in inputs, soil carbon dynamics, and accounting methods (Jordon et al 2022, Tadiello et al 2026). Many studies focus on individual practices rather than whole farming systems, with limited long-term evidence across diverse contexts, although this is beginning to change (Berthon et al 2025). Analyses often simplify yield or emissions outcomes using fixed yield penalties, yield-neutral assumptions, or uniform emissions factors—despite substantial variation across studies, systems, and locations (Pittelkow et al 2015, British Ecological Society 2025). Such simplifications are useful for exploring system-level outcomes, but should not be interpreted as representing RA as a single homogeneous production system. The growing prominence of RA in climate and environmental strategies therefore rests on a paradox: strong claims of system-wide benefits coexist with substantial uncertainty (British Ecological Society 2025). If regenerative systems maintain yields while reducing production emissions, they could make an important contribution to climate mitigation. However, even modest yield reductions may alter the balance of benefits. Food systems are globally connected, reduced local production is likely to be displaced (Grubb et al 2022), although how much will depend on changes in consumption, waste, and market dynamics. Consequently, local gains may be offset by additional land conversion and associated emissions elsewhere. The key question is whether RA can deliver net environmental benefits once any impacts on overall land demand are considered. Addressing this requires a system-wide perspective that explicitly accounts for production, land-use change, and displacement effects, alongside field-scale outcomes. 2. The system-level consequences of yield change Assuming demand remains constant, local production shortfalls must be met through expanded production elsewhere. This displacement of production can result in land use change in other regions, with associated GHG emissions and wider environmental impacts (Kastner et al 2021). This process, commonly termed displacement, negative spillover, or leakage, means that reducing emissions locally does not necessarily reduce total emissions, if lower yields require additional land to meet demand. Climate outcomes therefore depend not only on how land is managed, but also on how much land is ultimately required for production (Searchinger et al 2018). The effect of RA on both yield and production emissions is therefore critical. If RA reduces emissions while maintaining or increasing yields, total emissions may fall and land may be freed for ecological restoration, creating a potential ‘double climate dividend’ (Sun et al 2022). However, increased productivity does not necessarily result in an equivalent reduction in land under production, because lower production costs or prices can stimulate further production or expansion elsewhere (Phalan et al 2016, Balmford et al 2018). Conversely, if yields decline, emissions from displaced land-use change may counteract or exceed local emissions savings. The same intervention can therefore generate very different outcomes depending on its effect on productivity. Even modest yield changes can have substantial consequences when applied across large areas (Balmford et al 2018). Yet yield responses to RA remain highly uncertain, varying among crops, regions, baseline management systems, and stages of transition (Pittelkow et al 2015). Without robust evidence that yields are maintained, or explicit consideration of the consequences of yield change, RA climate impacts remain uncertain. 3. An illustrative land-use sensitivity analysis To explore how variation in RA yield and emissions outcomes influences system-level GHG balances, we developed a stylised land-use simulation representing a 10 000 ha UK arable landscape (figures S1 and S2 in supplementary methods). Yield changes from conventional management ranging from −25% to +10% were applied across the landscape, spanning the range of responses reported for RA and constituent management practices in the literature (supplementary methods; table S1). An upper limit for RA’s potential impact on crop production emissions was estimated from the best available evidence; we also considered a scenario with unchanged production emissions, reflecting uncertainty in implementation and effectiveness (supplementary methods). Although the model is UK-specific, the sensitivity analysis is intended to illustrate broader system-level relationships between productivity, emissions, and land-use. Our modelling of emissions captures only a subset of the practices associated with RA, focusing on reduced synthetic fertiliser use and cover-cropping effects. For simplicity, yield and emissions responses were treated as independent, although in practice the variables may be mechanistically linked; for example, changes in synthetic fertiliser may affect both emissions and yields. This approach allows a broad range of outcomes to be explored but may include combinations of yield and emissions responses that are less biophysically plausible than others. Where production fell below the conventional baseline, shortfalls were met through compensatory production allocated between domestic and international sources in proportion to their contribution to current UK supply chains, assuming that these sourcing patterns remain unchanged across scenarios. An additional sensitivity analysis assumed that all production shortfalls were met through overseas imports, representing a limiting case in which domestic production cannot expand to compensate for yield losses (figures S3 and S4). Additional production was assumed to occur either through conventional intensification of existing agricultural land or through extensification of conventional production onto new cropland. Under intensification, additional production generated production emissions but no land-use change emissions. In contrast, where production exceeded the conventional baseline, we assumed cropland was released for local habitat restoration, as reflected in UK land-use policy and restoration ambitions. Total GHG fluxes were calculated by summing emissions from domestic RA production, conventional domestic and international production, land-use change emissions arising from domestic and overseas extensification, and carbon sequestration associated with habitat recovery (supplementary methods). System-level land-use outcomes were highly sensitive to yield assumptions (figure 1). Because total production was assumed to be fixed, under extensification pathways, yield reductions under RA increased the amount of land required to satisfy demand, while yield gains reduced overall land requirements and created opportunities for habitat restoration. Under the largest yield penalty examined (−25%), total land requirements increased by up to 32%, whereas the largest yield gain (+10%) reduced land requirements by around 10%. Even modest yield changes produced substantial shifts in land demand across the production system, illustrating the sensitivity of land-use outcomes to assumptions about productivity. Climate implications depended on yield change, emissions reductions under RA, and the pathway through which compensatory production occurred (figure 2). Under intensification pathways (shaded yellow in figure 2), where additional demand was met on existing agricultural land, total emissions could decline even under moderate RA yield penalties, provided that meaningful reductions in production emissions under RA were realised. However, scope for achieving net climate benefits narrowed rapidly as yield penalties increased. Under extensification pathways (shaded blue in figure 2), where additional production required conversion of natural land to agriculture, outcomes were markedly different. In these scenarios, emissions associated with land-use change frequently outweighed reductions in production emissions, resulting in net increases in total system-wide emissions even assuming maximum reductions in production emissions under RA. Higher yield penalties generated larger increases in emissions because greater land demand meant more carbon losses from vegetation and soils. These results are illustrations, not predictions. Our deliberately simple model does not attempt to represent market mechanisms or the economic and behavioural responses that may influence how production shortfalls are accommodated; instead, it is intended to bound plausible system-wide outcomes. Likewise, the analysis focuses on GHG outcomes and does not quantify other potentially important ecosystem services of RA. Nevertheless, it demonstrates a fundamental principle: conclusions regarding the climate impact of RA depend critically on assumptions about yield outcomes and how production shortfalls are met. Small differences in yield can shift system-wide outcomes from net emissions reductions to net emissions increases, especially when displaced production occurs via extensification. Assessments of RA’s climate mitigation potential must therefore consider impacts on yield and hence land-use, alongside changes in production emissions. 4. Interpreting system-level trade-Offs Our scenarios depend on simplified assumptions about how land systems respond to changes in agricultural productivity. Yield changes are assumed to result in directly proportional changes in local production; and any area no longer needed for production (because of yield increases) is assumed to become available for habitat restoration. These assumptions provide useful boundary conditions, but neither is likely to hold perfectly in practice. Our assumption that production shortfalls lead directly to equivalent additional production elsewhere may overstate the extent of displacement. Agricultural systems operate within global commodity markets, where changes in supply influence prices, consumption, trade, and production decisions elsewhere (Bouyssou et al 2025). In practice, reduced output in one region may be partially offset through demand responses or substitution between commodities, with the magnitude of these responses determined in part by demand and trade elasticities. Elasticities vary across commodities, locations, and policy contexts, and therefore introduce uncertainty into estimates of displacement. These system-wide outcomes are not intrinsic to RA per se but emerge from how yield changes interact with land-use dynamics, market responses, and constraints on land-use. Yield gains likewise do not necessarily translate into proportional reductions in land under production. Rebound effects are typical, whereby greater yields tend to lower prices and/or increase profits, thereby stimulating increased production, and in some circumstances, agricultural expansion (Balmford et al 2018). Under such conditions, climate benefits associated with increased productivity are likely to be smaller than those estimated under ‘perfect’ land-sparing assumptions. However, rebound effects can be limited through interventions which conserve or restore natural habitats alongside stimulating yield increases (Phalan et al 2016). Land sparing could therefore occur where incentives and governance support both yield increases and habitat restoration. 5. Implications for research, policy and practice The illustrative analysis presented here highlights a fundamental challenge for RA: its climate impact cannot be evaluated independently of its effects on agricultural productivity. While regenerative practices may reduce production emissions and deliver wider environmental benefits, their overall impact depends on whether yields are maintained and how any production shortfalls are accommodated within the wider food system. Uncertainties about RA’s yield impacts, and how production is accommodated, are among the most influential determinants of system-level climate outcomes. Reported impacts of RA on yields and GHG emissions vary substantially across studies, while long-term evidence spanning different crops, climates, soils, and management systems remains limited (Pittelkow et al 2015, Berthon et al 2025, British Ecological Society 2025). Future analyses should seek to represent yield and emissions responses jointly, using empirical data or process-based models to capture relationships between management practices, productivity and GHG emissions. Uncertainty in production-emission reductions appears less influential, but there is still considerable uncertainty over the permanence of soil carbon gains and whether field-scale environmental benefits are realised at larger scales. Strengthening this evidence base across different contexts and interpretations of RA should therefore be a research priority. Our findings have implications for how RA is assessed and implemented. Analyses based on single assumed values for yield or emissions outcomes risk obscuring the range of possible system-level consequences. Instead, evaluations should consider uncertainty in productivity, emissions, land demand, and displacement, with scenario- and sensitivity-based approaches providing a practical means of testing the robustness of claimed benefits across plausible futures. RA should not be viewed as universally beneficial. Rather, its environmental performance depends on the balance between changes in productivity, production emissions, and land-use consequences across interconnected food systems. Assessments and policies that fail to account for these interactions risk overstating benefits or overlooking unintended impacts. System-wide evaluation will therefore be essential to understand whether RA can contribute credibly to climate change mitigation. Acknowledgment This work was supported by the Natural Environment Research Council (NERC) Changing the Environment Programme through their support for the Centre for Landscape Regeneration [Grant Number: NE/W00495X/1]. Data availability statement No new data were created or analysed in this study. Supporting Information available at https://doi.org/10.1088/1748-9326/aeac6c/data1.

About the Role This is a role for someone who knows that great campaigns don’t just inform people, they move them. Someone who can see the bigger picture, move quickly when an opportunity emerges,

PhD student Amanda Yee, working with Professor Lisa Dolovich, is exploring how to optimize medication care for older adults experiencing homelessness, with the goal of informing clinical practice and ...

Games are truly a powerful medium. They engage us and shape us. We play them, and in turn, they help inform how we interpret our own experiences. Games-based learning seeks to utilize all of these things. Game mechanics can engage users and help them solve problems.  Although in games, those problems are contrived challenges.