Incremental adaptation may be inadequate to deal with rapid shifts and tipping points for food production under climate change. The concepts of transformative and transformational adaptation have emerged in recent years to address the need for major, non-marginal transitions in sectors, such as agriculture in response to climate change. However, there is less empirical evidence of transformation in practice. Here we use a simple semi-quantitative definition to identify recorded cases of transformational adaptation in response to climate change. A structured search of academic literature found 23 empirical case studies that meet our criteria for transformation of agriculture under climate change: a response to climate risks along with a redistribution of at least a third in the primary factors of production (land, labor, capital) or the outputs and outcomes of production over a time period of 25 years or less. The case studies offer experience-based lessons on managing transformative processes in agriculture at all four stages of the adaptation cycle: understanding goals and objectives, developing a vision and pathway, implementing adaptation actions, and monitoring, evaluating and learning. In general, the case-study processes of transformation have diverged from well-managed, inclusive approaches based on foresight and continual learning. Our review provides little early evidence that transformative adaptation processes in response to climate change have generated more resilient agricultural systems or improvements in governance. Governments and development partners could improve the effectiveness of outcomes through providing more comprehensive and long-term approaches to adaptation planning alongside financial and technical assistance, within a framework that rewards farms as multi-functional systems.

The challenges facing agriculture in the coming decades are daunting. Recent research suggests that the impacts of climate change on agriculture and food systems may be wider-ranging than previously understood. Can feasible pathways to a food secure and sustainable future be identified? The scale of change required to meet the sustainable development goals, including those of no poverty, zero hunger and the urgent action needed to address climate change, will necessitate the transformation of local and global food systems. We identify eight elements of a theory of change to drive such transformation and highlight four pathways by which transformation may occur. We conclude with some suggestions for ‘business unusual’ for agricultural research for development.

Climate change impacts on agriculture have become evident, and threaten the achievement of global food security. On the other hand, the agricultural sector itself is a cause of climate change, and if actions are not taken, the sector might impede the achievement of global climate goals. Science-policy engagement efforts are crucial to ensure that scientific findings from agricultural research for development inform actions of governments, private sector, non-governmental organizations (NGOs) and international development partners, accelerating progress toward global goals. However, knowledge gaps on what works limit progress. In this paper, we analyzed 34 case studies of science-policy engagement efforts, drawn from six years of agricultural research for development efforts around climate-smart agriculture by the CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS). Based on lessons derived from these case studies, we critically assessed and refined the program theory of the CCAFS program, leading to a revised and improved program theory for science-policy engagement for agriculture research for development under climate change. This program theory offers a pragmatic pathway to enhance credibility, salience and legitimacy of research, which relies on engagement (participatory and demand-driven research processes), evidence (building scientific credibility while adopting an opportunistic and flexible approach) and outreach (effective communication and capacity building).

There is broad agreement that current food systems are not on a sustainable trajectory that will enable us to reach the Sustainable Development Goals by 2030, particularly in the face of anthropogenic climate change. Guided by a consideration of some food system reconfigurations in the past, we outline an agenda of work around four action areas: rerouting old systems into new trajectories; reducing risks; minimising the environmental footprint of food systems; and realigning the enablers of change needed to make new food systems function. Here we highlight food systems levers that, along with activities within these four action areas, may shift food systems towards more sustainable, inclusive, healthy and climate-resilient futures. These actions, summarised here, are presented in extended form in a report of an international initiative involving hundreds of stakeholders for reconfiguring food systems.

Science–policy engagement efforts to accelerate climate action in agricultural systems are key to enable the sector to contribute to climate and food security goals. However, lessons to improve science–policy engagement efforts in this context mostly come from successful efforts and are limited in terms of empirical scope. Moreover, lessons have not been generated systematically from failed science–policy engagement efforts. Such analysis using lessons from failure management can improve or even transform the efficacy of efforts. To address this knowledge gap, we examined challenges and failures faced in science–policy engagement efforts of the CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS). We developed an explanatory framework inspired by Cash et al.’s criteria for successful knowledge systems for sustainable development: credibility, salience, and legitimacy, complemented with insights from the wider literature. Using this framework in a survey, we identified factors which explain failure. To effectively manage these factors, we propose a novel approach for researchers working at the science–policy interface to fail intelligently, which involves planning for failure, minimizing risks, effective design, making failures visible, and learning from failures. This approach needs to be complemented by actions at the knowledge system level to create an enabling environment for science–policy interfaces.