SABERES TRANSDISCIPLINARES E ORGÂNICOS.

quinta-feira, 7 de maio de 2026

O líder certo para uma empresa deve refletir onde ela está no ciclo de vida, e não existe um protótipo único de "grande CEO" que funcione ao longo do ciclo de vida.


 Com o tempo, glorificamos construtores e conquistadores de impérios, tanto na política quanto nos negócios. Embora Steve Jobs tenha recebido status de lenda, por seu papel em trazer a Apple de volta dos mortos, o legado de Tim Cook na Apple, onde ele cumpriu com disciplina e contenção, merece respeito igual. O líder certo para uma empresa deve refletir onde ela está no ciclo de vida, e não existe um protótipo único de "grande CEO" que funcione ao longo do ciclo de vida.

O Índice de Liberdade Acadêmica de 2026 acabou de sair e não está nada bom... Deixe-me explicar:


 O Índice de Liberdade Acadêmica de 2026 acabou de sair e não está nada bom... Deixe-me explicar:

Todos os anos, 2300 especialistas medem a liberdade das universidades em 179 países. Eles avaliam se os pesquisadores podem se expressar livremente, escolher seu tema de pesquisa, colaborar com outros países, etc. Pois é, más notícias, as pontuações estão no pior 😞 nível de todos os tempos Cerca de cinquenta países regrediram, principalmente entre democracias ocidentais. O caso dos Estados Unidos, por exemplo, é bastante espetacular. Seu índice está diminuindo rapidamente, mais rápido do que o da Turquia ou da Hungria. Os EUA agora estão classificados em 85º lugar no mundo, atrás de países como Mongólia e Gana. A França também está recuando, mais discretamente. Essas não são proibições brutais como em um estado totalitário, mas sim uma pressão difusa sentida por pesquisadores em certas disciplinas. Por exemplo, alguns poderiam se censurar em assuntos sensíveis, por medo de não obter financiamento público. Resumindo, não são boas notícias, então...

Mais do que o dobro do volume de gás cortado devido ao fechamento efetivo do Estreito de Ormuz está sendo desperdiçado a cada ano


 Uau! Mais do que o dobro do volume de gás cortado devido ao fechamento efetivo do Estreito de Ormuz está sendo desperdiçado a cada ano porque os países não conseguem lidar com vazamentos de metano e queimas desnecessárias, diz o International Energy Agency (IEA).

COM DESTAQUE À BIODIVERSIDADE


 COM DESTAQUE À BIODIVERSIDADE


O Instituto Brasileiro de Geografia e Estatística (IBGE) divulgou um novo mapa-múndi com os continentes de cabeça para baixo e com o Brasil no centro. A publicação do “Riqueza de Espécies 2025” faz parte das celebrações de 90 anos do IBGE e em alusão ao Dia Internacional da Diversidade Biológica, comemorado em 22 de maio.

O presidente do IBGE, Marcio Pochmann, afirmou que “o novo mapa-múndi desafia séculos de visão eurocêntrica e reposiciona o Sul Global no centro do debate sobre biodiversidade, poder e futuro do planeta”.

Segundo o IBGE, “essa projeção busca promover uma visão mais justa e ‘descolonizada’ do mundo, corrigindo o viés eurocêntrico presente em mapas tradicionais e servindo como uma ferramenta educacional e de representação mais equilibrada”.

Unidades de Conservação e Terras Indígenas, organizei esses dados para visualizar sua distribuição no território nacional.


 Pesquisando sobre áreas preservadas no Brasil, especialmente Unidades de Conservação e Terras Indígenas, organizei esses dados para visualizar sua distribuição no território nacional.

No mapa, aparece uma presença em todo o país, mas com forte concentração na região Norte, com grandes áreas contínuas. O gráfico reforça esse padrão. Estados como Roraima 46,1%, Amazonas 29,4% e Pará 24,7% têm uma parcela muito elevada de seus territórios ocupada por Terras Indígenas, enquanto em grande parte do Sul, Sudeste e Nordeste essa proporção é bastante baixa. As duas visualizações juntas ajudam a entender não só onde estão essas áreas, mas também o peso que elas têm dentro de cada estado. Dados do Ministério do Meio Ambiente e do IBGE. Elaboração própria.

Eu sou a minhoca !


 O QUE EU REALMENTE FAÇO:

Como meu peso em matéria orgânica por dia. Folhas mortas, raízes em decomposição, bactérias, fungos — tudo que entra pela minha boca sai transformado. Minhas excretas (coprólitos) contêm 5 vezes mais nitrogênio, 7 vezes mais fósforo e 11 vezes mais potássio que o solo ao redor. Cada montículo enrolado que você vê na superfície do canteiro é adubo concentrado que eu produzi de graça. Escavo túneis de até 2 metros de profundidade. Cada túnel é um canal de aeração que leva oxigênio às raízes e um canal de drenagem que impede alagamento. Solo com minhocas infiltra água de chuva até 6 vezes mais rápido que solo sem. Aquele canteiro que alaga quando chove forte e seca em dois dias provavelmente não tem minhocas suficientes. Misturo camadas do solo. Carrego matéria orgânica da superfície para baixo e minerais de baixo para cima. Em um hectare, as minhocas movem entre 10 e 50 toneladas de solo por ano — o equivalente a uma camada de 5 mm de solo novo na superfície inteira. Seu canteiro está sendo lentamente reconstruído por baixo sem você perceber. TRÊS ESPÉCIES NO SEU QUINTAL: Minhoca-vermelha-da-califórnia (Eisenia fetida) — 6 a 8 cm, vermelha listrada. Vive nos primeiros 10 cm de solo e em composto. A estrela da compostagem doméstica. Não escava fundo — processa matéria orgânica na superfície. A que você compra para a composteira. Minhoca-gigante (Amynthas spp.) — até 15 cm, cinza-azulada iridescente. Se contorce violentamente quando perturbada. Escava túneis profundos. Muito comum em jardins do Sudeste. Produz coprólitos grandes e granulados na superfície. Minhoca-nativa (Pontoscolex corethrurus) — 8 a 12 cm, cinza-rosada. A minhoca mais comum em solos tropicais brasileiros. Encontrada em pastos, jardins e hortas. Tolera solos ácidos melhor que espécies exóticas. COMO LER O SOLO: Mais de 10 por pá: solo excelente, biologicamente ativo. Continue fazendo o que está fazendo. 5 a 10: solo razoável. Adicione composto e cobertura morta para alimentar mais minhocas. Menos de 3: solo debilitado. Provável compactação, falta de matéria orgânica ou uso de pesticida. Comece a adicionar composto na superfície sem revolver — as minhocas migram de áreas vizinhas em semanas se houver alimento. Zero: solo morto ou muito recente (canteiro novo com substrato comprado). Adicione composto e pare de usar qualquer pesticida de solo. As primeiras minhocas chegam em 1 a 3 meses. O QUE ME MATA: Pesticidas de solo — inseticidas granulados e fungicidas me matam por contato direto. Cada aplicação reduz minha população em 60 a 90%. Aração e revolvimento — destrói meus túneis, expõe meus ovos ao sol e me corta com as lâminas. Solo arado perde 70% das minhocas na primeira passagem. Solo descoberto — sem cobertura morta, o sol aquece a superfície a 50-60°C e me empurra para profundidades onde não consigo processar matéria orgânica. Excesso de sal de fertilizante químico concentrado.

A alfabetização no Brasil é feia, mas é uma flor.


 A alfabetização no Brasil é feia, mas é uma flor.

Entre 1934 e 1945, Carlos Drummond de Andrade foi chefe de gabinete no Ministério da Educação e da Saúde Pública e, nesse mesmo período, escreveu os livros A Rosa do Povo e Sentimento do Mundo. Absurdo, né? Entre os poemas dos dois livros está "A Flor e a Náusea", onde o poeta caminha por uma rua cinza e no meio do asfalto nasce uma flor feia e sem cor, mas nasce. Em 2025, o Brasil alfabetizou 66% das crianças, superando a meta do Compromisso Nacional Criança Alfabetizada. Dois anos atrás, eram 56%, e agora são cerca de 200 mil crianças a mais que estão alfabetizadas. É uma flor, o avanço é real, mas ela ainda é feia demais, pois um terço das crianças brasileiras das escolas públicas ainda não se alfabetizam na idade certa. Os resultados da alfabetização são divulgados nacionalmente e de maneira padronizada há apenas três anos. Essa divulgação ocorreu no mês passado, e em cima desses números produziram-se reportagens e rankings, e quem melhorou foi celebrado, quem não melhorou sentiu pressão. Para o bem ou para o mal, a possibilidade de comparar desempenhos incentiva a competição. Mas o papel de alguém que tenta analisar dados precisa ir muito além de rankear. Mais recentemente saíram os microdados, que mostram as informações escola por escola. Fui dar uma olhada e eis que tropecei em uma pedra no meio do caminho, com licença do Drummond. Os microdados revelam: 95% das escolas públicas têm pelo menos uma criança não alfabetizada. Noventa e cinco por cento. Quase todas. E quase metade das escolas públicas brasileiras tem entre 20% e 50% de não alfabetizados, sendo que mais da metade de todos os não alfabetizados estão nessas escolas. Esses números seguem propriedades de uma curva normal, não são uma grande surpresa para quem está acostumado com estatística, mas aqui aparece uma náusea. A náusea é que a atenção pública está voltada para um problema que tem outra forma. Hoje, manchetes, rankings e intervenções prioritárias atingem os melhores e piores do ranking. E não as “escolas do meio”, onde mora a maior parte do problema (mais da metade das crianças não alfabetizadas do nosso país). A pedra não é espetacular, não rende destaque, não recebeu atenção. Mas a pedra está ali, no meio do caminho. O analfabetismo invisibiliza, e a criança que não lê perde acesso ao código pelo qual o mundo se organiza. Quem não enxerga essa criança nos dados também reproduz essa invisibilidade, e as escolas do meio não ganharam manchete. Além disso, microdados ainda não trazem informações sobre raça/cor e educação especial, existem crianças que são ainda mais invisibilizadas. A náusea é dispersa, a pedra persiste, a flor precisa nascer. E todo mundo deveria poder ler Drummond.

Compartilhar mais sobre o que já estão fazendo para incentivar trabalhos orientados a sistemas.

 






  • Research Paper No systems transformation without systems literacy: Insights from CGIAR Hanna Ewell a,b,d,*,1, Eva Valencia Lenero c,d,e,1, Arwen Bailey f, Lennart Woltering a,g, Katharina Schiller h, Jon Hellin i a Knowledge, Technology and Innovation Department, Wageningen University and Research, the Netherlands b International Center for Tropical Agriculture (CIAT), Kenya c Systems Innovation, United Kingdom d Responsible Innovations, United States e Tricolor Coalition for Sustainability Transitions, Mexico f Bioversity International, Italy g International Center for Tropical Agriculture (CIAT), Tanzania h International Maize and Wheat Improvement Center (CIMMYT), Mexico i International Rice Research Institute (IRRI), Philippines H I G H L I G H T S G R A P H I C A L A B S T R A C T • Agri-food system transformation is needed to address polycrises and com plex challenges. • Systems approaches can help navigate complexity, addressing interdependent outcomes while minimizing trade-offs. • Yet there is low systems literacy, for utilizing systems metrics, and governing systems change. A R T I C L E I N F O Editor: Simon Fielke Keywords: Systems thinking AR4D Systems transformation Systems literacy A B S T R A C T CONTEXT: Agri-food systems face complex and intertwined social, ecological, economic and political challenges, that call for transformation. Systems approaches can help navigate these complexities by helping to sweep in perspectives, identify entry points to improve situations of concern, and minimize trade-offs. This requires systems literacy: the capacity to understand, identify and operationalize suitable systems approaches for specific problem types. OBJECTIVE: In this paper, we assess how CGIAR engages with and pursues systems thinking, as an example of a leading organisation driving change in agricultural research for development (AR4D), and explore the challenges for operationalizing systems thinking across the partnership. * Corresponding author. E-mail address: hanna.ewell@wur.nl (H. Ewell). 1 Lead authors. Contents lists available at ScienceDirect Agricultural Systems journal homepage: www.elsevier.com/locate/agsy https://doi.org/10.1016/j.agsy.2026.104662 Received 3 June 2025; Received in revised form 18 January 2026; Accepted 1 February 2026 Agricultural Systems 234 (2026) 104662 Available online 15 February 2026 0308-521X/© 2026 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). mailto:hanna.ewell@wur.nl www.sciencedirect.com/science/journal/0308521X https://www.elsevier.com/locate/agsy https://doi.org/10.1016/j.agsy.2026.104662 https://doi.org/10.1016/j.agsy.2026.104662 http://crossmark.crossref.org/dialog/?doi=10.1016/j.agsy.2026.104662&domain=pdf http://creativecommons.org/licenses/by/4.0/
  • METHODS: A qualitative approach was used combining a review of literature and semi-structured stakeholder interviews, ranging from current and past research staff, academics, and practitioners, to gather insights into the understanding and operationalization of systems thinking within the case study of CGIAR. RESULTS AND CONCLUSIONS: Our findings show that despite a commitment to food system transformation, to which systems approaches are inherent, systems thinking is underutilized in CGIAR. The results of the interviews point to limited systems literacy, with diverse interpretations of ‘systems thinking’, scattered adoption of systems approaches, and little intentional selection of different systems approaches for different problematic situations. They underscore an opportunity to better benefit from systems approaches by more intentionally engaging with various Systems approaches and fostering ‘systems literacy’, encompassing systems knowledge, governance and investment. SIGNIFICANCE: Systems thinking offers practical methodologies and tools to improve the complex and inter connected challenges towards agri-food system transformation, yet its effective application would benefit from greater ‘systems literacy’ in AR4D. 1. Introduction In a time of polycrises of increasing climate change, geopolitical turmoil, and food security risks, there is growing consensus among in ternational organizations and experts that a transformation of agri-food systems is necessary to achieve more sustainable and equitable out comes for both people and the planet (Béné and Abdulai, 2024). The potential for transformative change in current research and develop ment approaches is nevertheless largely constrained by locked-in structures divided in sectoral silos, with fragmented governance, and short-term funding priorities (Conti et al., 2021). Desirable in terdependencies are overlooked, opportunities for synergy are missed, and systems maintain mechanisms that have the risk of being both ineffective and inequitable (McGuire et al., 2025). The absence of in centives for cross-sector collaboration reinforces policy incoherence, and power imbalances that perpetuate inequalities (Amoak and Najiar, 2025). As a result, agricultural research for development (AR4D) often duplicates efforts, competes for resources, and fails to scale impact sustainably (Van Mil et al., 2014). While these shortcomings are well known, we highlight in this paper the potential of ‘systems literacy’ in overcoming them (Reynolds, 2011). Transforming agri-food systems requires moving beyond reduc tionist, sector-specific interventions towards approaches that recognize the interdependent, dynamic, uncertain, and multi-scalar nature of these systems (FAO, 2025; Van Mil et al., 2014; Conti et al., 2021; Klerkx and Begemann, 2020). Systems thinking offers ways to foster plural trans formation (Hall and Dijkman, 2019; Woltering et al., 2024; Röling, 2009; Leeuwis et al., 2021), making interconnections visible, fostering transdisciplinary partnerships, and amplifying diverse perspectives (FAO, 2025; Woltering et al., 2019). It facilitates “understand[ing] dy namics, so that interventions and investments can be better aligned to the social and political complexity and unpredictability of how human societies function and hence be more likely to have a positive impact” (Woodhill and Millican, 2023, p.6). Furthermore, it considers in terconnections among components and outcomes (FAO, 2025) and thus can help affected stakeholders to pinpoint areas or directions for expe dited and local action and agree on potential intervention points (Meadows, 2008; Chapman et al., 2022; Swinnen et al., 2024). Systems approaches provide avenues for action, making visible the complexity, high levels of uncertainty regarding what works, high disagreement on what should be done to solve a particular problem, and unpredictable feedback loops, which are also important properties of systems (FAO, 2025). By supporting joint reflection with relevant agri- food system stakeholders on the relationships, feedback loops, and emergent properties of food systems, systems approaches enable more coherent, inclusive, and adaptive responses. This is in line with Catalyst 2030's definition of systems change as “confronting root causes of issues (rather than symptoms) by transforming structures, customs, mindsets, power dynamics and policies, by strengthening collective power through the active collaboration of diverse people and organisations” (2024, p. 1). In this paper we explore the following research questions: i) To what extent have systems approaches been used in AR4D? ii) What barriers and opportunities exist to further implement systems thinking in AR4D practice? To answer these questions, we use a case study of CGIAR, a global research partnership for a food-secure future dedicated to transforming food, land, and water systems in a climate crisis. CGIAR is selected as case study as it has over 50 years of different engagements with systems thinking (from farming systems to food systems, using a variety of sys tems approaches). The paper is structured as follows: Section 2 outlines key roles of and types of systems approaches. This is followed in section 3 by an outline of the methodology applied and introduction to the case study of CGIAR, along with a historical perspective on the integration of systems thinking. In Section 4, an overview of systems approaches in CGIAR from the literature and perspectives from key informant researchers, academics and development practitioners are presented. The discussion, section 5, links the results with the concept of ‘systems literacy’ and recommendations for practical action in CGIAR as well as broader im plications for wider inclusion of systems literacy in AR4D are offered. Conclusions on the research findings and possible further research topics are offered in section 6. 2. Systems thinking and approaches Systems thinking refers to “all the various strands of thought and practice that make use of system philosophies, theories, perspectives, methodologies, models, models, methods, concepts and ideas to un derstand and intervene in the world” (Jackson, 2024, p xviii). These strands of thought and practice have emerged over decades from sci entific disciplines as wide as mathematics, anthropology, family therapy and complexity science (Ison, 2007). What they have in common is an action-oriented desire to “change a problematic situation for the better” (ibid., p. 143). Systems thinking recognizes different perspectives, relationships, feedback processes, and emergent properties (Leeuwis et al., 2021; Van Mil et al., 2014). It embraces an integrative and interdisciplinary viewpoint to recognize the diverse positive and negative factors influ encing change (Geels and Schot, 2007). It addresses ‘wicked problems’ to unlock progress across multiple objectives for sustained results and overcome the limits of siloed approaches to bring synergies and mini mize trade-offs (FAO, 2025). Thinking ‘systemically’ means focusing critically on the root eco nomic, social and political factors generated by structural patterns, along with the mindsets, beliefs, and worldviews that consistently in fluence the emergence of systems of interest and how they play out (Leach et al., 2012; Reynolds, 2024). Systems thinking recognizes that H. Ewell et al. Agricultural Systems 234 (2026) 104662 2
  • real-life situations are non-linear, are experienced differently by different stakeholders, and do not have clearly defined boundaries. This means recognizing that a change in one part of the system can trigger effects elsewhere or at a later time, that different stakeholders engage with the system for different reasons and pursue different goals, and that the boundaries we draw around the system are themselves socially constructed, and therefore open to being questioned and redefined. In other words, system ‘boundaries’ are determined by an observer and their perspective (see Fig. 1). For this reason, systems practitioners generally refer to a ‘system of interest,’ recognizing that it only exists to the extent that they construct it by identifying its parts and (inter)re lationships. Nobody can see a whole system; everyone identifies the elements they see as relevant, and these therefore create de facto ‘boundaries’ (Woodhill and Millican, 2023). Systems thinking is thus a philosophical, epistemological orientation towards understanding social realities and the actors embedded within them (Abercrombie et al., 2015). It goes beyond a technical methodol ogy. Its epistemology is characterized by an attentiveness to the circular and interdependent nature of phenomena, and an appreciation of how underlying structural arrangements shape the conditions that in dividuals and institutions face. Moreover, it foregrounds the reality that human actions frequently generate consequences that remain unfore seen or unacknowledged, underscoring the need for reflexivity and adaptive learning in both analysis and practice (Goodman, 1997). Critically, systems practitioners differentiate broadly between ‘thinking about systems’ (e.g. food, land and water systems, transport systems, sewage systems) as ‘real’ systems that are ‘out there’; and seeing the world ‘as systems’ (Fig. 1), as a device for exploring and improving dynamic problematic situations (Reynolds and Holwell, 2010; Gadsby and Wilding, 2024). Both traditions have relevance and significance, but one is more an ontological/hard tradition and the other is a more epistemological/soft tradition. For example, a ‘hard’ systems approach based on natural sciences proposes to understand, through time, the elements or drivers of change and relationships (Meadows, 2008). In comparison, a ‘soft’ systems approach involves more learning- oriented approaches that unpack assumptions and accept that “inquiry is never-ending” (Ison, 2007, p. 147). Ison (2007) links systems thinking and action research, showing that systems traditions provide conceptual tools, methodologies, and epistemological framings, enabling research to deal more effectively with complexity, uncertainty, multiple per spectives, and emergent change. This leads to the idea of systemic action research as a process of learning with and within systems, rather than simply applying systematic methods to predefined problems (ibid). Jackson (2024) makes the case for ‘systemic pluralism’, where practitioners select systems methodologies, or approaches, according to how they interpret the nature of the principal problem before them: • Mechanical approaches, rooted in engineering, focus on achieving a clearly specified goal that is defined outside the system itself. They treat the system as a set of components with inputs, outputs, coor dination and control mechanisms that can be optimized for efficiency and reliability. Systems Engineering is a typical example, where complex projects are decomposed into manageable parts, designed, tested and recombined to deliver predictable performance. • Inter-relational approaches emphasize the patterns of interaction, feedback loops and underlying structures that shape how a system behaves over time. Rather than focusing only on individual compo nents, they look at how reinforcing and balancing feedback generate emergent dynamics such as growth, resistance, delay or collapse. Systems Dynamics exemplifies this perspective, using causal loop diagrams and simulation models to explore how policy changes might unintentionally amplify or dampen key system behaviors. • Organismic approaches draw lessons from biology, viewing systems as living entities that adapt, co-evolve and self-organize in order to survive and thrive. They pay less attention to top-down command and control, and more to flows of information, learning, relation ships and the conditions that support resilience and regeneration. Sociotechnical Systems Thinking is one example, highlighting how Fig. 1. Seeing the world “as systems” with components, interrelationships and boundaries (Woodhill and Millican, 2023, p.11). H. Ewell et al. Agricultural Systems 234 (2026) 104662 3
  • social and technical elements must evolve together to sustain healthy organizations and communities. • Purposeful methodologies recognize that human action is organized through ‘human activity systems’, in which people come together around shared tasks but hold diverse roles, interests and worldviews. These approaches focus on making purposes explicit, surfacing contrasting perspectives and finding accommodations that allow collective action without pretending everyone fully agrees. Soft Systems Methodology is a prominent case, using tools like rich pic tures and conceptual models to help stakeholders jointly explore messy, ill-structured problem situations. • Emancipatory approaches ask those who shape or intervene in sys tems to critically examine whose interests are being served and who bears unacknowledged costs. They draw attention to unplanned ef fects on marginalized groups, the natural environment and future generations, and seek to rebalance power by amplifying excluded voices in analysis and decision-making. The Gender Equality, Envi ronments and Marginalized Voices evaluation framework (Stephens et al., 2018) is one example, providing structured guidance to interrogate equity, justice and sustainability within policies, pro grams and systems. 3. Methodology Drawing on both grey and peer-reviewed literature, we examined systems thinking definitions, boundaries and applications to form the conceptual basis. The literature search was carried out using Google Scholar, ScienceDirect, and Scopus, without restrictions on geography. It was not intended as an exhaustive or systematic review, since the objective was not to capture all available work on systems approaches, systems thinking and agri-food system transformation. Rather, publi cations were purposefully selected for their ability to illustrate key trends, and approaches across a broad body of literature. Using a case study analysis (Yin, 2017) of CGIAR, literature on historical and current applications, as well as insights from key informant interviews were used to examine the potential to operationalize systems thinking in AR4D building from past AR4D experiences, learnings, and current vi sions for agri-food transformation. 3.1. Case study CGIAR was selected as the case study for its global leadership in agricultural research for development (AR4D) and its relevance to the authors. Founded in 1971 to support national crop improvement and technology transfer, CGIAR now consists of 15 institutions working with more than 3000 partners across 90 countries to transform agri-food systems. Although CGIAR's impact is substantial—benefits are esti mated to be at least ten times its costs (Alston et al., 2020) —persistent challenges remain in translating research into sustainable change at scale. Since its inception, CGIAR has engaged with systems approaches in different ways. The integration of ‘systems-aware’ approaches such as Farming Systems Research in the 1970s and 80s, and Agricultural Knowledge and Information Systems (AKIS) in the 1990s laid important groundwork for engaging with systems complexity, seeing farms as whole systems, with farmers as actors, and combining biophysical and socioeconomic aspects (Byerlee et al., 1982). CGIAR centers began to adopt cropping and farming systems research as a way to move beyond single-variety or single-technology trials (FAO, 2001). The International Service for Na tional Agricultural Research (ISNAR) was established in 1979 explicitly to build capacity in responding to cross-sector demands, linking research organizations and stakeholders in a changing context, and learning for institutional innovation (Jannsen and Braunschweig, 2003). A decisive step came in the early 1990s, when the CGIAR formally expanded its mandate to include natural resource management (NRM) through the creation or incorporation of Centers with explicit NRM responsibilities—forestry (CIFOR), agroforestry (ICRAF/World Agro forestry), water (IWMI) and fisheries (WorldFish) (Renkow and Byerlee, 2010). This signaled recognition that productivity gains had to be reconciled with the sustainability of land, water, forests and aquatic systems, and that this required more integrated, cross-sectoral ap proaches (ibid). In the 2000s, Integrated Natural Resources Management (INRM) was promoted as a way to pursue systems- and landscape-level research, with frameworks that stressed adaptive management, stakeholder participation and cross-scale integration (ICRAF, 2000). Reviews of NRM and systems research in this era already highlighted issues that remain current: difficulties in measuring impact of integrated systems approaches; the need for stronger theories of change; and the challenge of working effectively across disciplines, sectors and institutions (Independent Science and Partnership Council (ISPC), 2011). 2010 saw the design of ‘integrated systems’ CGIAR research pro grams (CRPs) - Humidtropics, Dryland Systems, and Aquatic Agricul tural Systems (AAS) – each explicitly adopting different systems approaches. For example, AAS used Participatory Action Research to collaboratively define research priorities, while Humidtropics applied the Multi-Level Perspective framework to guide systemic change (Wigboldus and Leeuwis, 2013). However, several systemwide reviews noted persistent challenges: integrated systems projects were often seen as diffuse, hard to manage, and difficult to evaluate, especially compared to more focused commodity research. Leeuwis et al. (2017) characterized systems research in CGIAR as “an arena of struggle”, in which competing discourses about the role of research (upstream sci ence vs. embedded research-for-development), the nature of evidence, and acceptable forms of partnership played out. Systems approaches were championed as necessary for complex, real-world problems, but were also criticized as fuzzy, expensive and risky in terms of producing publishable science and clearly attributable impacts (ibid). Conse quently, these programs were discontinued within two years (Douthwaite and Hoffecker, 2017; Holderness et al., 2021; Leeuwis and Wigboldus, 2017). A synthesis of experiences from these research pro grams suggests that this discontinuation was a missed opportunity for advancing systems transformation within CGIAR (Douthwaite and Hoffecker, 2017). From 2019 onward, CGIAR began explicitly advancing “trans formation of food, land and water systems in a climate crisis” (CGIAR System Organization, 2021, p. 17), bringing attention back to systems. Nevertheless, while “systems transformation” was one of the three major research Action Areas (ibid), there was limited clarity and standards on what was meant by ‘systems’, ‘transformation’ or the combination thereof (Woltering, 2024). Despite these repeated efforts to integrate systems research since the 1980s, and several examples of participatory and even emancipatory approaches, organizational models and donor frameworks typically still favor projects with well-defined, short timelines, clear resource boundaries, and impact assessed in terms of the number of farmers adopting a particular innovation. Social science approaches remain underrepresented, with biophysical approaches dominating AR4D agendas (Barrett et al., 2022; McGuire et al., 2024). This explains the dominance of a reductionist tendency and, hence, an engineering - or “hard”- take on systems (Hall and Dijkman, 2019; Haynes et al., 2020; Klerkx et al., 2012; Leeuwis et al., 2017, 2021; Woltering, 2024). The current competitive funding context has exacerbated this tendency, evolving over the decades from long-term, center-based models to short- or medium-term, project-driven programs, increasingly shaped by donor priorities (Beintema and Echevarria (2020)). This has led to research fragmentation and a focus on rapidly scalable, visible innovations, often at the expense of integrated, transformative science (Lynam et al., 2024; Woltering et al., 2024). Recognition of the limits of ‘scaling innovation’ models as sufficient to transform food, land and water systems has led to a focus within CGIAR on achieving outcomes through innovation use at scale, referred H. Ewell et al. Agricultural Systems 234 (2026) 104662 4
  • to as outcome-oriented scaling (Schut et al., 2020), with the suggestion that this is an approach to foster the conditions for transformative change. The Scaling Scan (Jacobs et al., 2018) and Scaling Readiness (Sartas et al., 2020) similarly attempt to expand scaling approaches to address context, stakeholder engagement, and the evolution of skills and capacities needed to coordinate scaling efforts, i.e. complementing the mechanical, scaling perspective, with a more purposeful approach, bringing in different stakeholders around shared desired outcomes rather than the innovations as an end in themselves (Woltering et al., 2024). The current CGIAR portfolio (CGIAR, 2025) comprises nine syner gistic Science Programs, which with the help of four accelerators and building on a “robust foundation of over 50 years of history of impact and thousands of ongoing projects,” aim to support “addressing major global challenges such as climate change, gender and social inequalities, poor-quality diets, rural poverty, environmental degradation, and issues stemming from fragility, conflict, and violence” (CGIAR, 2025). A dedicated “Scaling for Impact” program has been designed for scaling inclusive and responsible innovation benefits for transformative impact (ibid). 3.2. Interviews A total of 19 semi-structured interviews were conducted online and in person between February and July 2024. Notes were recorded digi tally for online interviews, and handwritten notes taken for in-person interviews, and later transcribed. They were then collated, after which codes were generated deductively (Fife and Gossner, 2024) on the key barriers and opportunities identified. Interviewees were selected based on their experience and interest in promoting systems thinking within CGIAR. They represented researchers, funding partners, consultants and academics engaged in systems thinking and practice and in agriculture/ food systems (see Table 1 below). We used snowball sampling as a purposeful approach suitable for “locating information-rich informants” (Patton, 2002, p237). This may introduce bias inherent to purposeful selection (Naderifar et al., 2017) and a lack of diversity in responses, as respondents may move in similar circles. However, it is a valuable strategy when information-holders are not known across a population, and where the objective of the research is to gain expert depth on an issue rather than representing the opinions of a range of actors across AR4D. The interviews shed light on the processes, opportunities, and challenges of implementing systems thinking and practice in CGIAR. This allowed us to assess the current and potential applications of sys tems approaches in the wider AR4D context. See Annex 1 for the list of interview questions. 4. Results: CGIAR's use of and engagement with systems approaches 4.1. What approaches have been used in CGIAR? A review of literature revealed that CCIAR has been engaging in a wide variety of systems approaches since the 1980s (see also section 3.1 in the results above), categorized according to Jackson's five method ologies (Jackson, 2024). 4.1.1. Mechanistic approach A typical mechanistic systems approach applied by CGIAR is the Rogers' innovation adoption curve (1983), which has shaped the design of development interventions, the framing of impact evaluations, and decisions about new investments in AR4D (Glover et al., 2019). The predominance of this diffusion-adoption model is strongly linked to the legacy of the Green Revolution, which fostered a narrative centered on technical change, scaling, and the expectation that widespread dissem ination of useful innovations would lead to positive system-wide impacts (Van Etten, 2022). There is widespread use of LogFrames and Theories of Change, which map out a project's inputs, outputs and outcomes and impact in a linear fashion (Belcher et al., 2024). Through participatory systems mapping, these can more explicitly align with the wider context and account for potential negative or unexpected outcomes (Wilkinson et al., 2021). Initial scaling approaches tended to be in this model, based on the premise that an innovation proven effective at small scale will yield proportionally greater benefits at larger scale and impact viewed primarily as a function of reach (Woltering et al., 2024). 4.1.2. Interrelational approach From an interrelational perspective, Systems Dynamics tools such as causal loop diagrams have been widely used in CGIAR. These tools offer a holistic approach to problem-solving that considers how the parts of a system interact with one another (Spielmann et al., 2019) to create a ‘range of possible outcomes’ (Meadows, 2008 in Leeuwis et al., 2021 p.762). Causal loop diagrams have been used to explore the dynamics of situations ranging from tropical tree seed systems (Valette et al., 2020), to the relationship between climate action and peacebuilding (Moralez- Munoz et al., 2022), to understanding climate and health (Tepa-Yotto et al., 2024), among others. By incorporating positive and negative feedback loops of relationships within agri-food systems, and by iden tifying underlying drivers of change, this approach has potential to facilitate the development of transformative, inclusive actions aimed at fostering positive sustainable societal change (McGuire et al., 2024; Amoak and Najiar, 2025). By bringing in reflexivity, it also encourages greater adaptive capacity to emerging conditions and stakeholder needs and interests (McGuire et al., 2025). Another way to use interrelational lenses has been using place-based methodologies, and socio-technical and socio-environmental analysis. These lenses have used causal loop diagrams to better reflect and analyze the system structures and interactions in a given place. For example, relationship mapping tools were tailored to the purpose of accommodating needs of local practitioners, policymakers or farmers within crop breeding efforts (Govaerts et al., 2021). Also, there have been proposals to group innovations in socio-technical innovation bundles (Barrett et al., 2022). Rather than scaling single solutions, these programs pursue inclusive bundling that includes training, access to credit and other complementary approaches required from the context. These types of approaches have been used in large bilateral programs such as the Accelerating Impacts of CGIAR Climate Research for Africa (AICCRA), to more effectively scale climate-smart agriculture (CSA) and Climate Information Services (CIS) (Ewell et al., 2025). 4.1.3. Organismic approach A seminal framework for analyzing systems change in agri-food systems, from an organismic perspective, is the Multi-Level Table 1 Interviewees affiliation. ID Affiliation 1 International Livestock Research Institute (ILRI) 2 Commonwealth Scientific and Industrial Research Organisation (CSIRO) 3 International Food Policy Research Institute (IFPRI) 4 International Food Policy Research Institute (IFPRI) 5 Wageningen University and Research (WUR) 6 Open University, UK and Monash University, Australia 7 Open University, UK 8 Ex-International Center for Maize and Wheat Improvement (CIMMYT) 9 French National Institute for Agricultural Research (INRA) 10 CGIAR Systems Organisation 11 Alliance Bioversity-CIAT 12 Ex-International Rice Research Institute (IRRI) 13 Ex-World Fish 14 Wageningen University and Research (WUR) 15 German Development Corporation (GIZ) 16 Long-time CGIAR consultant 17 World Bank 18 CGIAR Systems Organisation 19 Wageningen University and Research (WUR) H. Ewell et al. Agricultural Systems 234 (2026) 104662 5

  • Na minha carreira, deixei de focar principalmente em dietas saudáveis porque continuava vendo o progresso ser desacelerado por decisões fragmentadas. Queria focar mais em como as decisões #FoodSystems levam em conta as interconexões.


    Mas não sou ingênuo quanto à realidade: existem desincentivos e riscos reais em agir de forma mais integrada.

    Por isso, este artigo recente sobre as barreiras e oportunidades para operacionalizar #systemsapproaches em CGIAR realmente ressoou comigo. Mostra que, apesar do forte compromisso com a transformação de sistemas na CG, o pensamento sistêmico continua subutilizado, e mesmo aqueles que dizem trabalhar com sistemas frequentemente o fazem de forma restrita.

    Por quê? Porque os pesquisadores não têm incentivos para trabalhar além das fronteiras. Duas razões principais emergem das descobertas do artigo:

    🤕 Primeiro, abordagens sistêmicas raramente produzem os impactos claramente atribuíveis necessários para financiamento e publicação. Se o impacto é definido como mudança de sistemas, ele demora a emergir e é difícil de medir exatamente o que as métricas atuais não recompensam.
    🥵 Segundo, a pesquisa baseada em sistemas é mais difícil em termos de gestão. Isso exige abranger diferentes disciplinas e temas, além de todo o tempo e esforço que isso implica. Nas palavras do artigo, é "mais borroso" e "difuso". Sem recompensas, faz sentido racional evitar abraçar a complexidade.

    O que este artigo realmente mostra é que, na realidade cotidiana dos pesquisadores, abordagens sistêmicas os expõem a riscos. O triste fato é que, enquanto as abordagens de sistemas carregarem riscos profissionais e institucionais maiores do que permanecerem em silos, elas permanecerão marginais, independentemente de quantas vezes pedimos "transformação de sistemas".

    Instituições de pesquisa, financiadores e editoras poderiam fazer muito para mudar isso – e compartilhar mais sobre o que já estão fazendo para incentivar trabalhos orientados a sistemas. Adoraria ouvir exemplos concretos.

    𝐌𝐨𝐝𝐞𝐥𝐨𝐬 𝐝𝐞 𝐫𝐞𝐠𝐫𝐞𝐬𝐬𝐚̃𝐨 𝐜𝐨𝐦𝐨 𝐫𝐞𝐝𝐞𝐬 𝐧𝐞𝐮𝐫𝐚𝐢𝐬

     



    🛠️ Talvez a ferramenta mais usada da estatística sejam os 𝐌𝐨𝐝𝐞𝐥𝐨𝐬 𝐝𝐞 𝐑𝐞𝐠𝐫𝐞𝐬𝐬𝐚̃𝐨.
    🤖 Esses modelos estabelecem uma ponte entre a estatística clássica e os modelos preditivos usados em 𝐚𝐩𝐫𝐞𝐧𝐝𝐢𝐳𝐚𝐝𝐨 𝐝𝐞 𝐦𝐚́𝐪𝐮𝐢𝐧𝐚.

    📊 Por meio deles, podemos compreender relações entre variáveis, avaliar a importância de diferentes fatores e realizar previsões para novos casos.
    📉Ao contrário do que muitas vezes se imagina, 𝐨𝐬 𝐦𝐨𝐝𝐞𝐥𝐨𝐬 𝐝𝐞 𝐫𝐞𝐠𝐫𝐞𝐬𝐬𝐚̃𝐨 𝐧𝐚̃𝐨 𝐬𝐞 𝐥𝐢𝐦𝐢𝐭𝐚𝐦 𝐚 𝐫𝐞𝐥𝐚𝐜̧𝐨̃𝐞𝐬 𝐥𝐢𝐧𝐞𝐚𝐫𝐞𝐬.

    🚀 Eles também podem ser usados para modelar relações não lineares entre variáveis, seja por meio de transformações, termos polinomiais, funções não lineares ou modelos mais flexíveis, como regressão logística, regressão spline e modelos aditivos generalizados.

    🧠 De fato, alguns modelos de regressão podem ser vistos como casos particulares de 𝐫𝐞𝐝𝐞𝐬 𝐧𝐞𝐮𝐫𝐚𝐢𝐬. Por exemplo, a regressão linear corresponde a uma rede neural sem camadas ocultas e com função de ativação linear.
    🤖 De modo semelhante, a regressão logística pode ser interpretada como uma rede neural simples, também sem camadas ocultas, mas com uma função de ativação sigmoide na saída.

    🚀 Por isso, compreender esses modelos é fundamental para quem trabalha com dados, estatística, aprendizado de máquina e deep learning.