Research

INTRODUCTION

The question of how we can manage our planet’s resources to produce enough healthy food without destroying our life-support system has been raised in the discussion around the biodiversity-production mutualisms (Seppelt et al. 2020). Agricultural intensification, i.e., landscape homogenization (fewer, larger fields), and increasing inputs (labor, fertilizer, irrigation, pesticides), has led to an increase in food production in many parts of the world (Levers et al. 2018, Seppelt et al. 2020). Although this is important to sustain increasing human populations and for food security, it also leads to a degradation and/or collapse of ecosystem functions such as soil fertility, pollination, and biocontrol (Carpenter and Brock 2008, Peterson et al. 2018, Seppelt et al. 2020). This degradation of agroecosystems is probably the cause of the decline in yields in around 23% of cultivated landscapes (IPBES 2019). Despite strong intensification of agriculture globally, around 9% of the human population suffers from hunger, showing a global gap between overnutrition and famine (FAO 2023). An imbalance of power relations between global and national decision making and those who are dependent on agricultural goods contributes to a trade-off between yield and ecosystem functions that leads to degradation and still fails to create food security. Power relations can be described by the concept of agency as “the freedom people have to negotiate their own lives (including their own resilience) in face of adverse circumstances” (Béné et al. 2012:12). Projections show that the concentration of agricultural land into large farms increases with economic growth (Lowder et al. 2021). This concentration can lead to unequal distributions of agricultural products, not only within national markets but also globally, as global value chains often perpetuate power imbalances (Lang et al. 2023). This inequity in distribution goes hand in hand with the fact that processes of transitions to more sustainable food systems are at the nexus of power imbalances between innovators and those guarding the stability of an existing system (Wezel et al. 2020). In this sense, the power in the agricultural landscape lies with those who promote unsustainable farming practices throughout the control of markets and industries (e.g., fertilizer and pesticide industries), disregarding the fact that the agricultural landscape is a common good that must be cared for in order to provide a good livelihood for future generations.

To address the issues arising from (i) agricultural intensification and (ii) the imbalance of power relations, it is necessary to integrate ecosystem functions and biodiversity as well as social and economic interactions into agricultural system thinking. The concept of resilience, i.e., agroecological resilience as described in the next section, has the potential to provide transformation paths in the face of current ecological, social, and economic obstacles in agriculture. Unfortunately, the term resilience is understood differently because of its use in many different disciplines (Carpenter et al. 2001, Peterson et al. 2018, Martin et al. 2019), which makes it even harder to develop a comparable resilience assessment (Quinlan et al. 2016, Serfilippi and Ramnath 2018, Dakos and Kéfi 2022). Generally resilience refers to the capacity of a system, to maintain or restore its functions, structure, and general identity under changing environmental influences (Walker et al. 2004, Folke et al. 2010, Martin et al. 2019). Recently, the assessment of social-ecological resilience has been examined in various studies (Kéfi et al. 2014, Standish et al. 2014, Allen et al. 2016, Quinlan et al. 2016, Baho et al. 2017, Ingrisch and Bahn 2018, Dakos and Kéfi 2022). In agroecosystems, resilience is discussed as a concept to develop potentially practical, comparable metrics for system behavior in the face of disturbances, such as drought and landscape-related management. Thus, it can serve as a monitoring tool to prevent agroecosystem degradation and loss of ecosystem functions. Implementation of resilience can lead to a sustained balance between productivity and functionality in agroecosystems (Peterson et al. 2018). Agroecological resilience as part of an intact system of ecosystem functions (Oliver et al. 2015) is a fundamental prerequisite for a more comprehensive resilience of the interlinked agri-food sector, as is becoming apparent from Vialatte et al. (2019) and Ewert et al. (2023). However, there is also criticism of the concept. In addition to a multitude of definitions in various disciplines and the resultant varying of understanding of the term resilience (Cretney 2014, Oliver et al. 2018), there are doubts about the possibility to actually quantify resilience in a comparative way (Quinlan et al. 2016, Dakos and Kéfi 2022). The main obstacle is the dynamic nature of resilience, which is characterized by many different processes within the system itself, but also by the larger system context (cross-scale dynamics). In addition, measuring social-ecological resilience often requires the joint development of indicators with local actors, practitioners, and knowledge holders because of its context dependency (Dakos and Kéfi 2022, UNU-IAS 2024). When considering resilience, the individuals who influence the system are often excluded, which should be avoided, to prevent the term from becoming meaningless. This was the case, for example, with the concept of “sustainable development” as pointed out by Béné et al. (2012). The local reference to stakeholders, however, can limit cross-case comparisons (Quinlan et al. 2016) and thus prevent the goal of using resilience measurements as a large scale monitoring tool. Measuring resilience requires simplifying complex systems, i.e., setting boundaries, assuming a closed system, distinguishing disturbances, selecting a limited number of resilience metrics, or using existing data instead of the data actually needed (Dakos and Kéfi 2022). To ensure the quality of future resilience research study designs, we refer to the six recommendations for the application of resilience assessment and management developed by Quinlan et al. (2016).

In current approaches to assess agroecological resilience, the development of indicators has been used broadly, rather than direct measurement (Billeter et al. 2008, van Oudenhoven et al. 2011, Cabell and Oelofse 2012, Martin et al. 2019). To ensure holistic consideration of the nature of complex systems and the concept of resilience, it is essential to base measurements of resilience on a broad set of indicators to ensure a deeper understanding of the system (Quinlan et al. 2016, Dakos and Kéfi 2022). In agroecosystems, several studies summarize indicators of resilience:

• Indicators of resilience in social-ecological production landscapes and seascapes (SEPLs; UNU-IAS 2024),
• Agroecosystem resilience indicators (Córdoba et al. 2020, Tittonell 2020),
• Indicators of functional resilience based on measures of biodiversity and other indicators of functional resilience (Martin et al. 2019),
• Indicator to study crop resilience in agroecosystems to stress events (Peterson et al. 2018),
• Behavior-based indicators for assessing agroecosystem resilience (Cabell and Oelofse 2012), and
• Social-ecological indicators of natural and agrarian landscape resilience (van Oudenhoven et al. 2011).

The overall aim of our systematic review is to expand resilience research by providing a practical proposal for a feasible quantitative measurement of agroecological resilience from a landscape perspective. Although, we have pointed out the importance of power relations, a thorough analysis of these relations is beyond the scope of this study. We therefore implicitly examine the power relation on the basis of three common resilience concepts (ecological, social, and economic resilience). This research focuses on three questions to investigate how agroecological resilience has been assessed in practice thus far: (i) What types of research approaches and resilience concepts have been used to assess agroecological resilience? (ii) What resilience capacities can be assigned to the studies to assess agroecological resilience from a landscape perspective? (iii) How were the research approaches and resilience concepts used to demonstrate and assess resilience capacities? In doing so, we contribute to the recommendations described by Quinlan et al. (2016) for the application of resilience thinking in management activities: (i) basing ongoing efforts on quantitative resilience assessments on theory, (ii) understanding the specific context of resilience measurement by asking framing questions (Resilience for what purpose? Resilience of what? Resilience to what and over what time frame? Resilience for whom?), and (iii) scale-sensitive conceptualizing of resilience.

We start by thoroughly defining agroecological resilience along the framing questions of resilience. Below we describe the methodological approach used to systematically review the assessment of agroecological resilience. Because there is already a lot of conceptual work on resilience assessment, i.e., system and indicator frameworks on agroecosystem resilience (Peterson et al. 2018, Martin et al. 2019, Córdoba et al. 2020, Tittonell 2020), we based our systematic review on a literature search for recently published assessments of the agroecosystem behavior from a landscape perspective that draw heavily on the concept of resilience. The exact procedure for selecting the articles to be included in the review analysis and the statistical methodology is described in the methods section. We emphasize the statistical results by describing the agroecological resilience approaches of the core articles that emerged from the statistical analysis. In the discussion, we point to gaps in the measurement of agroecological resilience and propose concrete solutions for a practical agroecological resilience measurement.

Agroecological resilience

We propose a definition of agroecological resilience based on the definition of Folke et al. (2010): “the capacity of a social-ecological system to continually change and adapt yet remain within critical thresholds.” Resilience can be described by three resilience capacities: robustness, adaptation, and transformation (Meuwissen et al. 2020; Table 1). Because resilience is one of the 10 elements of agroecology defined by the FAO (2018), it lends itself to look at resilience from an agroecological perspective. Agroecology is defined as “an integrated approach that simultaneously applies ecological and social concepts and principles to the design and management of food and agricultural systems” (FAO 2018:1). Alongside sustainable intensification, agroecology is being discussed and implemented as a solution to the current challenges in the food system, not only for low-income smallholders but also for industrialized agriculture (Boix-Fayos and Vente 2023, Ewert et al. 2023). Therefore, we define agroecological resilience in the sense that we look at the capacity of the holistic agroecosystem (including practical, social, and economic aspects) to respond continuously and dynamically to disturbances. The agroecological system can withstand disturbances (robustness) or continually adapt or transform in response to disturbances (adaptability or transformability; Table 1). The critical thresholds that define the agroecological resilience are described by a multitude of agroecological resilience indicators that must be maintained for the system to remain or transform into an agroecological system (e.g., Córdoba et al. 2020, Tittonell 2020, UNU-IAS 2024).

We look at possible agroecological resilience assessments from a landscape perspective. Resilience is part of the third level of agroecological transformation (Ewert et al. 2023) defined by Gliessman (2016:188): “redesign the agroecosystem so that it functions on the basis of a new set of ecological processes.” To redesign agroecosystems, interlinkages between different ecological and social processes and associated challenges must be taken into account. A landscape perspective on agroecosystems enables the assessment of resilience, taking into account the relationship between land use, biodiversity, and different ecosystem functions, at several levels (Kozar et al. 2023). A landscape-scale agroecological resilience assessment approach could contribute to the transformation of agroecosystems. Transformation can be achieved by the redesign of agroecosystems through the possibility of identifying obstacles in resilience enhancement and by introducing a comparable monitoring tool that is carefully scaled-up from local to landscape levels.

Methodological approach

To investigate the assessment of agroecological resilience we developed an analytical framework considering the framing questions of resilience: Resilience for what purpose (Quinlan et al. 2016)? Resilience of what (Carpenter et al. 2001)? Resilience to what and over what time frame (Carpenter et al. 2001, Oliver et al. 2018)? Resilience for whom (Cretney 2014; Fig. 1)?

Resilience of what?

Our study focused on the resilience of agroecosystems from a landscape perspective. A landscape approach allows the following:

• Integrating different components of production (i.e., ecosystem services, ecological processes, biodiversity, agricultural management, and social structures),
• Understanding the interactions between stakeholders (e.g., primary production, food supply chains, domestic transport networks, and households),
• Explaining trade-offs between prioritizations, and
• Exploring and implementing transformative governance paradigms in agroecosystems (Vialatte et al. 2019, Kozar et al. 2023).

Biophysically, we focused on agroecosystems and their boundary elements from a landscape perspective, i.e., at a level larger than individual farms but small enough to consider local environmental conditions, such as weather or species distributions, and local social conditions, such as management practices or yield. It allows to investigate what factors and indicators have been incorporated into the assessment of agroecological resilience undertaken by the reviewed studies.

Resilience for what purpose?

The purpose of resilience measurement can be described by the respective resilience focus of the study examined, which we explain by the three resilience capacities (Table 1), as extensively discussed by Meuwissen et al. (2020).

Resilience for whom?

In order to include socioeconomic and cultural aspects in our investigation of agroecological resilience measurement and to ensure the most comprehensive analysis of agroecological resilience and its interactions, we incorporated social and economic resilience concepts, and vulnerability, always in the context of ecological resilience. The concept of vulnerability focuses on resilience based on system weakness instead of system strength (Béné et al. 2012, Serfilippi and Ramnath 2018).

Even if our interest in strengthening resilience relates in particular to the intrinsic functionality of agroecosystems (i.e., ecological resilience), agricultural landscapes are under pressure from at least two perspectives that cannot be examined separately, i.e., ensuring agricultural production (social and economic interests) and provision of ecosystem services (ecological interests; Seppelt et al. 2020). The diversity of perspectives on agricultural landscapes leads to totally different agroecosystems depending on human interest, values, and criteria of management practices (Cretney 2014, Córdoba et al. 2020). For example, farmers who are economically dependent on agriculture have a different view of and interest in agriculture than those who are not directly financially dependent on it, but who may make political decisions about how the land can and should be farmed (Suškevičs et al. 2023). Therefore, agroecosystems are defined not only by their ecological resilience but also by symbolic, economic, political, and technological relationships at different scales and power hierarchies (Cretney 2014, Vialatte et al. 2019, Córdoba et al. 2020). Because it is more likely that qualitative assessments reflect different perspectives on agricultural landscapes, we have included them in our review in an effort to ensure a holistic understanding of resilience in agroecosystems.

Resilience to what and over what time frame?

It is necessary to examine which disturbances the system must be able to withstand. In agroecosystems, a mosaic of repeated and different disturbances appears at different levels in space (e.g., landscape-related management or global climate change) and time (e.g., recurring acute pesticide application or chronic habitat loss; Peterson et al. 2018, Martin et al. 2019). Some of these disturbances are predictable, others are not, leading to a large uncertainty in the assessment of resilience (FAO 2021).

We synthesize the types of disturbances considered for agroecological resilience assessment in the reviewed studies. We did not focus on a specific disturbance or time frame, but summarized data structure (i.e., data collection, data input, spatial and temporal levels) in order to provide an overview to the question “Resilience to what and over what time frame?”

METHODS

To conduct a reproducible systematic review, we followed the PRISMA 2020, a guideline for systematic reviews to transparently report why the review was done, what the authors did, and what they found (Page et al. 2021). First, we identified eligibility criteria: article type, focus, ecosystem, and spatial scale (Table 2). Second, we systematically searched the research databases Scopus and ISI Web of Knowledge. Third, we collected and screened the identified records, including the use of machine learning for process validation. After coding, we analyzed and synthesized the coded information following our analytical framework (Fig. 1).

Review process

Search strategy

To analyze agroecological resilience assessment from a landscape perspective we included articles considering ecological aspects, e.g., biodiversity or ecosystem functions of agricultural landscapes. We considered articles worldwide to provide a full overview on assessments. We screened the identified reports of the last 20 years (from 2003) because the main reference models in resilience analysis have emerged since 2008 (Serfilippi and Ramnath 2018), including a five-year buffer. Exclusion criteria for the screening process are summarized in Table 2.

We searched peer-reviewed articles in Scopus and ISI Web of Knowledge on 19 May 2022. Based on a previous review (van der Lee et al. 2022) about the resilience of social-ecological systems, we developed a search string for title, abstract, and keywords. We considered studies with different resilience concepts and did not limit it to one definition. This was necessary because few approaches to practical resilience measurement exist to date. The search string included the different foci that ecological resilience can have in agricultural landscapes and the different possibilities to describe assessment of agroecological resilience. The complete string, the resulting articles, and the review protocol can be found in Appendix 1.

Selection process

The selection process was carried out by two reviewers, a main reviewer and one for verification purposes to ensure an objective review process. After excluding duplicate articles and articles of an incorrect type (see “article type” in Table 2), 820 articles were screened by the first reviewer for the eligibility criteria. Five percent of these articles were checked individually by the second reviewer (n = 42). We used ASReview as an additional instance for further validation of the screening process to ensure that all important articles were included into full-text screening. ASReview uses screening prioritization by rearranging the articles to be reviewed. The software learns from the reviewers’ decisions and selects with this learned information the next articles presented to the reviewer (ASReview LAB developers 2023). To train the artificial intelligence software using the default settings, 826 articles were read into ASReview. The screening process in ASReview ended when 40 articles in a row were deemed irrelevant. The whole screening process resulted in 165 articles assessed for eligibility. In the end, 42 articles were included into the analysis. An overview on the selection process is provided in Appendix 1 (for more detailed information, see the link provided in the Data Statement, https://doi.org/10.5281/zenodo.15077447).

Data collection process

We created a simple database to collect data of the whole articles, mainly of the methodological approaches (please see the Data Statement). Most of the coding followed a deductive approach, using categories from the relevant literature. Variables describing limited or diverse knowledge were coded inductively. The combination of deductive and inductive coding resulted in a mixed data structure of qualitative and quantitative variables (Table 3). The coding was performed by the first reviewer and validated by the second reviewer.

Statistical analysis

The statistical analysis of the coded results of the reviewed studies (n = 42) was performed in three steps:

First, the coded qualitative variables were aggregated using a word frequency analysis (R-package tm). It reduced the qualitative data to the most frequently used word stems (e.g., “management” was reduced to “manag”). Punctuation and white space were automatically deleted. We removed sparse terms, and kept those terms that appear in more than 10% of the qualitative variables (threshold = 0.9). The cosine distance was then calculated for the remaining main terms, with which the terms were weighted in relation to the frequency of their use. The result of the word frequency analysis combines the word frequency of each qualitative variable in all coded articles and the relative importance of the entire dataset of reviewed articles by criteria (Ferreira et al. 2013).

In the second step, multiple factor analysis was performed (R-package FactorMineR) because this can take into account the mixed structure of our data. This allowed us to examine both the internal consistency within the groups and the way in which the different groups relate to each other across all variables, both quantitative and qualitative. The multiple factor analysis reduced the complexity of the data and highlighted similarities between the reviewed articles (Lê et al. 2008). Variables that took into account the research approach (data structure, research approach, resilience concept, disturbance, and resilience capacity) were used for axis construction because analysis of actual agroecological resilience assessment approaches is the key interest of this review. Variables describing the system under study (origin of the system and system description), and the analysis (indicators, factors, and scoring method) were implemented as supplementary variables because these variables describe what was actually assessed as part of the applied research approach. Assessment versus research approach refers to the overall view of resilience vs. the actual research design.

In the final step, hierarchical cluster analysis was used to determine optimal clusters (R-package FactorMineR). This allowed us to analyze which types (i.e., clusters) of approaches to assess the behavior of agroecological systems distinguished the reviewed articles, and to examine what arguments could be made about the system by using each approach. The cluster number with the highest relative gain was chosen for grouping reviewed articles with similar criteria (Lê et al. 2008). In order to underpin the statistical results with qualitative examples directly from the articles reviewed, we visually identified core articles from the center of the “Factor map” (please see the Data Statement).

The results of the hierarchical cluster analysis were visualized in radar-charts (R-package fmsb). In addition, UpSet-plots were created (R-package ComplexHeatmap) to visualize the intersections of several quantitative variables. The R-Markdown can be found in the Data Statement.

RESULTS

The hierarchical cluster analysis resulted in two clusters. The most important variables for constructing dimensions one and two were data structure and research approach. The first cluster included 27 of the 42 articles analyzed, while the second cluster included 15 articles. The multiple factor analysis was able to explain 19.5% of the variation by the first, and 10.7% by the second dimension. The word frequency analysis resulted in an approximate appraisal of the results of disturbances, factors, and indicators of agroecological resilience (Appendix 2).

The two clusters of the hierarchical cluster analysis can be described in terms of the research approaches used (Fig. 2A). The blue cluster represents modeling approaches combined with case studies and spatial analysis (GIS-based or remote sensing for the analysis of, e.g., vulnerability or landscape, population, climate or water dynamics). We will therefore refer to this cluster as the quantitative cluster further on. The yellow cluster represents qualitative approaches (e.g., questionnaires, interviews, stakeholder workshops) in combination with case study approaches, so we will refer to this cluster as the qualitative cluster. We summarized the results of the core articles of the two clusters of our hierarchical cluster analysis in Figure 3, to provide an overview on actual agroecological resilience assessments. Nine of 42 articles mixed the two approaches (Figs. 2B and 3), one of which also represented was also a core article of the qualitative cluster (i.e., Ticktin et al. 2018). Most mixed articles were part of the qualitative cluster (Fig. 3).

How is resilience assessed?

The quantitative cluster measured indicators mainly with weighted indices, and/or mathematical analysis or used a predetermined computation of indices, e.g., assessment of vulnerability of plant species by quantifying exposure, sensitivity and adaptive capacity (Crossman et al. 2012). In the qualitative cluster, the characteristics were measured primarily through the scoring of different categories and through perceptual judgments by observers or interviewees, e.g., stakeholder judgement about local ecological knowledge (Ticktin et al. 2018). Only a few articles of the qualitative cluster included mathematical analyses and computations in their assessment approaches (Figs. 3 and 4). For example, Badmos et al. (2015) simulate soil loss using multi-agent simulations and Lee et al. (2021) use an in-situ data acquisition process of participatory rapid rural appraisal and agroecosystem analysis to investigate the biophysical and human factors that interact in time and space and affect social-ecological resilience in rangelands.

Resilience to what and over what time frame?

Particularly in the qualitative cluster, the articles did not mention specific disturbances to which the system should be resilient. Apart from that, no pattern could be derived from the distribution of disturbances across both clusters (Appendix 2).

When assessing agroecological resilience, differences in the data structure used were apparent (Fig. 5). The quantitative cluster used predominately secondary data, e.g., economic data from government sources (Carper et al. 2021). The qualitative cluster used a combination of primary and secondary data and also used more primary data than the quantitative cluster (Fig. 5A), e.g., data derived from interviews (Ticktin et al. 2018, Vanwindekens et al. 2018). In both clusters, combinations of spatial and non-spatial data were used for the analysis, although the quantitative cluster used spatially explicit data more frequently, e.g., data about geolocated species occurrence (Crossman et al. 2012), water table depth (Carper et al. 2021), or crop rotation (Gardner et al. 2021). The qualitative cluster used mainly non-spatial data (Fig. 5B), e.g., stakeholder surveys (Castonguay et al. 2016, Ticktin et al. 2018, Lee et al. 2021). The spatial scale of the analysis in both clusters was mostly between the local level and the biome to landscape level. This is most likely due to the fact that we focused our review on studies applied to the landscape level. However, the quantitative cluster referred more often to the larger level than the qualitative cluster (Fig. 5C), i.e., biome-landscape (Tittonell et al. 2006, Carper et al. 2021, Gardner et al. 2021). The temporal scale of data analysis in the quantitative cluster spanned across multiple years (e.g., Carper et al. 2021, Gardner et al. 2021), while in the qualitative cluster the temporal scale of the data was often not explicitly stated (e.g., Ruiz-Agudelo et al. 2015, Vanwindekens et al. 2018; Fig. 5D). However, when the time frame was specified, the qualitative cluster also referred to multiple years (e.g., Lee et al. 2021).

Resilience for whom and for what purpose?

All studies focused on ecological resilience (Figs. 3 and 6A), which corresponded to our eligibility criteria. The qualitative cluster completely included social resilience in addition to ecological resilience and also considered economic resilience more frequently, although these concepts were not queried by our search string. Generally, the qualitative cluster targeted a broader scope of resilience than the quantitative. Three core articles from the qualitative cluster dealt exclusively with ecological resilience, i.e., a focus on hydrological environmental flow requirements (Hashemi et al. 2022), support and stabilization of pollinator populations and services (Gardner et al. 2021), and impacts on soil caused by land-use (Tittonell et al. 2006). Four of the qualitative studies included all concepts of resilience (i.e., Ruiz-Agudelo et al. 2015, Castonguay et al. 2016, Vanwindekens et al. 2018, Lee et al. 2021). Ruiz-Agudelo et al. (2015) integrated different system components, i.e., climate change scenarios for the region, agricultural management for reducing land degradation rates, government policy that encourages sustainable agricultural production systems and rural development, and the effects of changes in the price for agricultural and livestock products. Castonguay et al. (2016) considered bio- and agro-diversity, ecosystem functionality, traditional knowledge, and economic security. Vanwindekens et al. (2018) and Lee et al. (2021) were able to integrate requirements from practice by involving stakeholders from the onset in the scientific process.

All resilience concepts used in the two clusters referred to robustness as a resilience capacity (Figs. 3 and 6B). The qualitative cluster, with a focus on social resilience, was able to reflect adaptability more strongly than the quantitative cluster. Transformability could be addressed exclusively by the quantitative cluster, specifically, two quantitative core articles. Crossman et al. (2012) represented the only core article that considered all three resilience capacities (Fig. 3), i.e., robustness as the vulnerability of plant species to climate change, adaptability as the species’ ability to migrate, and transformability as a single priority index to identify priority areas for species conservation, which is a radical proposal for change in management. By studying historical land use and cover changes, Marull et al. (2019) investigated transformability as past socio-metabolic transition periods characterized by changes in energy storage and information. The results of the transformative momentum are used to make suggestions for the future management of agricultural land to overcome social-ecological degradation, i.e., the implementation of low external input strategies based on innovative enhancements of rural cultural knowledge that can empower farming communities in the market and the public sphere.

Resilience of what?

The system properties showing assumptions about the system under assessment can be described by the agricultural system description and the data structure (Table 3). The agricultural systems considered can be described by the agricultural system and the agricultural management class (Table 3). There were hardly recognizable patterns distinguishing the clusters for the agricultural system class (Figs. 3 and 7A). The articles in both clusters mainly assessed crops and diverse cultivation systems, the qualitative cluster had a slightly stronger focus on these systems and additionally on grassland and pastures. About 25% of both clusters did not specify the class of agricultural system, with the quantitative cluster integrating more studies without specification at this point. In contrast, a clear pattern emerged for the agricultural management class (Figs. 3 and 7B). More than a half of the articles in the quantitative cluster did not specify the management practices of the systems evaluated. The remaining studies of the quantitative cluster investigated primarily rotational systems, mixed cropping systems, intensive and extensive systems. Similar to the quantitative cluster, more than 25% of the qualitative cluster did not specify the management, but those that did, compared intensive to extensive management or assessed the agroecological resilience of mixed cropping systems and extensive management systems. We could not find a clear correlation in terms of the agricultural system or management used in the core articles.

We found no major differences between the clusters in the features used to assess agroecological resilience (Appendix 2). Combining these results with the assessment methods (Fig. 4) and data structure results (i.e., Fig. 5B), we can suggest a dual use of factors and indicators for resilience assessment: first, a spatially explicit and quantitative assessment of landscape metrics in the quantitative cluster, and second, a non-spatial qualitative assessment in the qualitative cluster of components related to management practices and the respective influences on agroecological resilience.

DISCUSSION

Our results show a research gap in quantitative measurements of agroecological resilience. Social aspects could be better integrated into the modeling of agricultural landscapes to ensure a holistic representation of agroecological resilience. Only the quantitative approaches took into account the transformability capacity of resilience. Combined with a lack of social context in quantitative measurements, this points to an important gap in the assessment of transformability, in which social aspects should be a prerequisite to represent agroecological resilience or vulnerability in a holistic way (Cretney 2014, Allen et al. 2016, Oliver et al. 2018, Schlüter et al. 2022). The quantitative articles assessed agroecological resilience in a spatially explicit way, in contrast to the qualitative research, which most likely depends on the more quantitative scoring methods used. The qualitative studies did not specifically assess agroecosystem behavior to a defined disturbance that the studied system must withstand or the time frame considered in the evaluation, while articles of the quantitative cluster lacked a definition of the agricultural management in the respectively studied system. Qualitative studies were based on a broad use of resilience concepts, while the quantitative cluster was mainly based on ecological resilience. Generally, the transformability and adaptability of agroecological resilience has been less studied than robustness. Future potential for agroecological resilience assessment lies in (i) assessing the resilience of agroecosystems in all aspects of social-ecological complexity, and (ii) capturing the transformability potential of agroecosystems.

Gaps in holistic resilience assessment

One important system assumption about resilience is for whom resilience is assessed (Cretney 2014), i.e., the resilience concept. Our results supported Córdoba et al. (2020) stating that quantitative assessments often do not consider power relations. By including social context and power relations, not only the capabilities of the local systems can be considered but also the structures that determine these capabilities can be defined in relation to each other. The assessment of resilience should not be reduced to the ability to access technology or biophysical resources, but should also include human well-being (Córdoba et al. 2020).

The three core articles that included only the ecological resilience concept (Fig. 3; i.e., Tittonell et al. 2006, Gardner et al. 2021, Hashemi et al. 2022) provided good approaches to examine ecological thresholds, but failed to consider the social and economic context. The purely ecological perspective in these papers should be broadened by jointly developing management practices with local farmers as key stakeholders. These studies would gain value by incorporating information on management options, including recognizing interactions of global trends and their consequences for pluralistic diversity (i.e., diversity in actors, ecosystems, production systems, nutrients, values, knowledge, narratives, livelihoods, scales, and institutions), connectivity and system feedbacks (Bennett et al. 2021, Kozar et al. 2023). The recommendations resulting from the models developed in the three core articles would therefore not be mere recommendations, but practical, actionable advice for a change in agricultural management that ensures ecosystem functions.

The absence of the social sphere in the studies of the quantitative cluster was also reflected in the fact that almost 75% of the papers did not mention the type of management practice of their study area and thus did not sufficiently consider the direct human impact (Fig. 7B). The proposed focus on landscape metrics and land cover as indicators of agricultural management in the quantitative cluster articles points to the underestimated influence of social dynamics on landscape patterns. Although some management practices can be analyzed using landscape metrics, such as field size, boundary features, or crop rotation (e.g., Gardner et al. 2021), other major human influences that manifest themselves in certain management practices, such as pesticide or fertilizer use, can be insufficiently captured using landscape metrics. A specification of management that goes beyond simple classifications like extensive vs. intensive or organic vs. conventional management would help ensure a holistic indication of agroecosystem resilience.

Our results suggest that the qualitative cluster may represent a more holistic view on agroecological system behavior than the quantitative cluster, however, less precisely calculated (Fig. 4). Of the qualitative core articles, the approaches of Lee et al. (2021) and Ruiz-Agudelo et al. (2015) show how quantitative agroecological resilience assessment can be embedded into local contexts and needs. The qualitatively and quantitatively mixed approaches allowed them to discuss problems with stakeholders and develop joint solutions, which led to improved knowledge transfer and long-term effects with regard to possible adaptation processes. This kind of outcome is an advantage of the landscape approach (Kozar et al. 2023). Such approaches have the potential to reveal power relations because there is generally a high level of participation throughout the research process. A possible extension to the approach of Lee et al. (2021) could be to specify different management practices in a given area in order to investigate the specific results experimentally.

After considering resilience concepts used in both clusters, we analyzed which resilience capacities were addressed by them. It appears that resilience is often used to adapt system states to the current yield-based economic system, which generally maintains inequalities (Cretney 2014, Córdoba et al. 2020). But as Lee et al. (2021:139) phrase it for the country of Mexico, “It is a tough reality that rural development policies meant to improve well-being and standard of living among poor populations are unsustainable because they produce land degradation, rangeland disequilibrium, and water insecurity so severe that resources become insufficient to sustain livelihood.” Although adaptation might not be the resilience capacity that should be pursued, transformability might offer the potential to actually improve biodiversity and ecosystem functions of agroecosystems by reshaping agricultural production under equitable structures to ensure a good quality of life for all involved stakeholders (Cretney 2014, Kozar et al. 2023). The inadequate consideration of transformability in the papers reviewed, is likely due to the multiple challenges in governing the transformation to sustainable food systems. This includes the current lack of information on the impact of governance systems on sustainable transformation, including a diversity of actors, values, and narratives on the distribution of landscape benefits and ecosystem services (Kozar et al. 2023). Some of those radical changes are certainly difficult to imagine in advance and therefore difficult to incorporate into practical quantitative research. Modeling studies have the potential to analyze various pathways of social-ecological system development under defined conditions, which can contribute to the development of an integrative understanding and action to increase resilience (Schlüter et al. 2019). The development of future scenarios with different focal points could identify obstacles, trade-offs, and interdependencies between ecological, social, economic, and institutional values and interests in order to improve the resilience of agroecosystems in a transformative way. Crossman et al. (2012) and Marull et al. (2019) examined the transformability based on historical data in order to suggest possible implementations for the future. However, the extent of transformability assessed in both were vague as a broader transformative approach would also include opportunities for transformational processes to translate the proposed management options into real action, including exploring the impacts on different stakeholders, from more marginalized groups to institutions associated with power (Kozar et al. 2023). Another reason for the lack of focus on transformability in resilience assessment could be that transformative processes often take place outside scientific research processes. Scientific evidence may be incorporated into management practices, but these implementations are often not scientifically monitored and generally not published in peer-reviewed journals (Cretney 2014).

Gaps in the development of mixed methods approaches

One important advantage of the quantitative cluster compared to the qualitative cluster was that the represented data was spatially explicit (Fig. 5B). The approach of Carper et al. (2021), for example, resulted in heatmaps, depicting levels of resilience metrics in their study area for market inflation shock and canal supply shock, and thus visualizing areas of high and low resilience. The development of different spatially explicit scenarios could enable the exploration of possible system development pathways that offer the potential to investigate how single or multiple changes in relevant parts of the system may influence vulnerability (Ruiz-Agudelo et al. 2015) and strengthen resilience. Additionally, spatially explicit scenarios offer the potential to incorporate disturbances and time frames into resilience assessments. The approach of Carper et al. (2021) provides a good opportunity to measure resilience by including two different disturbances (market inflation shock and canal supply shock). When it comes to the actual assessment of resilience, consideration of disturbances is indispensable, as the behavior before and after a disturbance determines resilience by definition. The development of mixed-method approaches is promising for capturing the complex interplay of ecological, social, and economic factors in agroecological resilience assessment.

Suggested directions of future agroecological resilience assessment

Although realistic quantification of agroecological resilience will not be possible in the near future due to a lack of theoretical knowledge about power structures in the agroecosystem (Cretney 2014, Córdoba et al. 2020), we see potential in a two-step approach to enable approximate quantification of agroecological resilience: (i) prior spatially explicit modeling, and (ii) practical experiments at landscape scale.

Prior spatially explicit modeling of agroecological resilience

Suitable indicators for agroecological resilience should first be identified in cooperation with key stakeholders at the landscape level under consideration (cf. Vanwindekens et al. 2018, Lee et al. 2021). Indicators and possible disturbances would be identified in collaboration with stakeholders and placed in the system context in stakeholder workshops, using cognitive mapping, group model building, or similar methods (cf. Ruiz-Agudelo et al. 2015). We suggest the combined use of the indicators yield and biodiversity (e.g., ecosystems species, traits) or ecosystem functions (e.g., nutrient cycling, antagonist abundances) for agroecological resilience measurement because they reflect the biodiversity-production mutualisms (Seppelt et al. 2020). The indicators identified can be categorized into the four drivers of ecological resilience (Peterson et al. 2018): crop diversity, landscape diversity and connectivity, biological community dynamics, and resource and micro-environmental regulations. To include the social dimension, the listed indicators on ecological resilience can be expanded by the indicators from Córdoba et al. (2020): political-organizational factors, use of resources, land tenure, production relationships, biophysical factors, social factors, health factors, and soil management and biodiversity. We propose to follow our analytical framework by carefully and precisely answering and documenting the framing questions of agroecological resilience.

Once the local indicators and their interdependencies have been identified, the quantified values can be fed into system dynamics models, state-and-transition models, Markov chain models, Bayesian Belief Networks, or other models that can describe relationships and feedbacks between variables (cf. Schlüter et al. 2019). The measurement of agroecosystem resilience could be improved by integrating the qualitative approaches into the quantitative modeling approaches, possibly leading to a more precise calculation of social-ecological variables and spatially explicit modeling capabilities. Quantification can be achieved through literature research, expert interviews, and stakeholder participation. We also suggest adding the agricultural boundaries of the most important pressures on biodiversity (cf. García-Vega et al. 2024) as possible threshold values in the calculation of agroecological resilience. In a final step, the chosen model can be coupled with a spatially explicit model, as is possible for Bayesian Belief Networks, for example. The output could be a map of potentially high and low resilient agricultural landscape areas under different scenarios (cf. Carper et al. 2021, Milazzo et al. 2022). Again, it is important to note that such an approach would constitute a first indication of agroecological resilience but not a sufficient evaluation. To validate the model results of agroecological resilience assessment, it would be useful to design and carry out agroecological resilience monitoring at the landscape level for a long time period to discover changes in resilience.

Practical experiments at landscape scale to assess agroecological resilience on the ground

A scientifically monitored practical experiment offers the opportunity to explore how resilience can be monitored and evaluated. An approach that takes into account power structures in decision making can only be achieved through local collaborative approaches that prioritize the needs of the people who inhabit and manage a landscape and must precede the final monitoring and evaluation of resilience (Córdoba et al. 2020). We therefore advocate starting resilience measurements with local people at the landscape scale. This can be done, for example, by carrying out an indicator assessment, following UNU-IAS (2024), the TAPE approach outlined by Mottet et al. (2020) and Tittonell (2020). In all cases, local land managers are involved from the onset. In addition, the approaches enable comparable quantification, which can be used for longer-term monitoring. Without this fundamental understanding of social contexts, resilience monitoring and evaluation can lead to uncertain results and incorrect implementation of measures.

A collaborative, inter- and transdisciplinary holistic social-ecological research project based on the prior involvement of local stakeholders and investigating various resilience-enhancing processes can be established as a landscape approach in regions with a particular need to increase resilience. A focus on significant turning points of resilience, e.g., in the transformation from an intensively farmed agricultural system to an extensively or agro-ecologically managed system, could allow a better understanding of the relationship between resilience and agricultural management practices. We refer directly to agroecological practice because agroecology, by definition, is based on a holistic, equitable view of the agricultural system (FAO 2018, Ewert et al. 2023). This means that power structures can be broken down through the implementation of agroecological practices. As already mentioned, resilience is one of the transformation stages toward agroecology (Gliessman 2016, Ewert et al. 2023), which is why it is only logical to consider agroecology in the establishment of practical experiments to measure agroecological resilience. Such an approach requires long-term funding, as the monitoring and evaluation of resilience can only take place after the occurrence of disturbance.

CONCLUSION

We found a broad range of possible methods for assessing agroecological resilience. Not all of the studies examined were designed to capture agroecological resilience in all its complexity, as we also included studies that did not use resilience as a guiding concept. Nonetheless, the multitude of different approaches provided us with a valuable glance of possible approaches that can be combined to approximate a quantification of agroecological resilience. We found that a particular potential for improvement lies in (i) assessing the resilience of agroecosystems in their high social-ecological complexity, and (ii) capturing the transformability potential of agroecosystems. Current research on agroecological resilience fails to initiate a transformation process toward more resilient agroecosystems and thus a more resilient food system.

Because we focused on the assessment of agroecological resilience, we were unable to analyze the consideration of power hierarchies in all its depth. This was beyond the scope of our study and remains a future interdisciplinary field of research in resilience thinking. However, we were able to gather evidence on the consideration of power hierarchies by integrating social, economic, and ecological resilience concepts and our suggestions for improvement have great potential to ameliorate future agroecological resilience approaches at the landscape level. For future assessment approaches of agroecological resilience, we suggest focusing on a collaborative and participative approach that particularly includes the opinions and views of those who manage the land and directly depend on agricultural goods, i.e., farmers and the local community. Although we advocate landscape-based approaches, we would like to point out how important it is to consider the global interdependencies in food supply chains. The transformation of agricultural landscapes at the landscape level in one place will have an impact on global market distribution. Especially when it comes to food security, these interdependencies need to be taken into account. There is already a lot of literature that deals with the practical implementation of agroecological transformation of food systems, which we would like to refer to for further details (Wezel et al. 2020, Ewert et al. 2023, Niggli et al. 2023, Schiavo et al. 2023).

Regarding our research questions, we found the following:

1. The promising possibility to combine quantitative modeling with qualitative assessment of local stakeholder needs. This allows for an agroecological resilience assessment that considers all resilience concepts (i.e., ecological, social, and economic resilience).
2. A large gap in the assessment of transformability in comparison to a sufficient assessment of robustness and adaptability.
3. The indication that only modeling approaches were able to account for transformability. Thus, modeling approaches in agroecological resilience assessments are promising to expand the notion of transformation processes in agricultural landscapes with the aim of improving agroecological resilience.

The most important requirements for initial agroecological resilience modeling can be summarized in one approach: a participatory, spatially explicit analysis of the agricultural landscape resilience, taking into account power relations and feedbacks between humans and the environment. In this way, all three reflections about system assumptions described by Kozar et al. (2023) can be met. The participation of key stakeholders could be used to set transformation targets, design concrete experiments, and build a working knowledge for resilience assessment.

LITERATURE CITED

Allen, C. R., D. G. Angeler, G. S. Cumming, C. Folke, D. Twidwell, and D. R. Uden. 2016. Quantifying spatial resilience. Journal of Applied Ecology 53(3):625-635. https://doi.org/10.1111/1365-2664.12634

ASReview LAB developers. 2023. ASReview LAB: Active learning for systematic reviews. https://asreview.readthedocs.io/en/latest/index.html

Badmos, B., S. Agodzo, G. Villamor, and S. Odai. 2015. An approach for simulating soil loss from an agro-ecosystem using multi-agent simulation: a case study for semi-arid Ghana. Land 4(3):607-626. https://doi.org/10.3390/land4030607

Baho, D. L., C. R. Allen, A. S. Garmestani, H. B. Fried-Petersen, S. E. Renes, L. Gunderson, and D. G. Angeler. 2017. A quantitative framework for assessing ecological resilience. Ecology and Society 22(3):17. https://doi.org/10.5751/ES-09427-220317

Béné, C., R. G. Wood, A. Newsham, and M. Davies. 2012. Resilience: new utopia or new tyranny? Reflection about the potentials and limits of the concept of resilience in relation to vulnerability reduction programmes. IDS Working Papers 2012(405):1-61. https://doi.org/10.1111/j.2040-0209.2012.00405.x

Bennett, E. M., J. Baird, H. Baulch, R. Chaplin-Kramer, E. Fraser, P. Loring, P. Morrison, L. Parrott, K. Sherren, K. J. Winkler, J. Cimon-Morin, M.-J. Fortin, B. L. Kurylyk, J. Lundholm, M. Poulin, J. T. Rieb, A. Gonzalez, G. M. Hickey, M. Humphries, K. Bahadur KC, and D. Lapen. 2021. Ecosystem services and the resilience of agricultural landscapes. Pages 1-43 in D. A. Bohan and A. J. Vanbergen, editors. Advances in ecological research. Academic, London, UK. https://doi.org/10.1016/bs.aecr.2021.01.001

Billeter, R., J. Liira, D. Bailey, R. Bugter, P. Arens, I. Augenstein, S. Aviron, J. Baudry, R. Bukacek, F. Burel, M. Cerny, G. de Blust, R. de Cock, T. Diekötter, H. Dietz, J. Dirksen, C. Dormann, W. Durka, M. Frenzel, R. Hamersky, F. Hendrickx, F. Herzog, S. Klotz, B. Koolstra, A. Lausch, D. Le Coeur, J. P. Maelfait, P. Opdam, M. Roubalova, A. Schermann, N. Schermann, T. Schmidt, O. Schweiger, M. Smulders, M. Speelmans, P. Simova, J. Verboom, W. van Wingerden, M. Zobel, and P. J. Edwards. 2008. Indicators for biodiversity in agricultural landscapes: a pan-European study. Journal of Applied Ecology 45(1):141-150. https://doi.org/10.1111/j.1365-2664.2007.01393.x

Boix-Fayos, C., and J. de Vente. 2023. Challenges and potential pathways towards sustainable agriculture within the European Green Deal. Agricultural Systems 207:103634. https://doi.org/10.1016/j.agsy.2023.103634

Cabell, J. F., and M. Oelofse. 2012. An indicator framework for assessing agroecosystem resilience. Ecology and Society 17(1):18. https://doi.org/10.5751/ES-04666-170118

Calleros-Islas, A. 2019. Sustainability assessment. An adaptive low-input tool applied to the management of agroecosystems in México. Ecological Indicators 105:386-397. https://doi.org/10.1016/j.ecolind.2017.12.040

Carpenter, S. R., and W. A. Brock. 2008. Adaptive capacity and traps. Ecology and Society 13(2):40. https://doi.org/10.5751/ES-02716-130240

Carpenter, S., B. Walker, J. M. Anderies, and N. Abel. 2001. From metaphor to measurement: resilience of what to what? Ecosystems 4(8):765-781. https://doi.org/10.1007/s10021-001-0045-9

Carper, J. M., M. R. Alizadeh, J. F. Adamowski, A. Inam, and J. J. Malard. 2021. Quantifying the transient shock response of dynamic agroecosystem variables for improved socio-environmental resilience. Ecology and Society 26(2):17. https://doi.org/10.5751/ES-12354-260217

Castonguay, A. C., B. Burkhard, F. Müller, F. G. Horgan, and J. Settele. 2016. Resilience and adaptability of rice terrace social-ecological systems: a case study of a local community’s perception in Banaue, Philippines. Ecology and Society 21(2):15. https://doi.org/10.5751/ES-08348-210215

Córdoba, C., C. Triviño, and J. Toro Calderón. 2020. Agroecosystem resilience. A conceptual and methodological framework for evaluation. PLoS ONE 15(4):e0220349. https://doi.org/10.1371/journal.pone.0220349

Cretney, R. 2014. Resilience for whom? Emerging critical geographies of socio‐ecological resilience. Geography Compass 8(9):627-640. https://doi.org/10.1111/gec3.12154

Crossman, N. D., B. A. Bryan, and D. M. Summers. 2012. Identifying priority areas for reducing species vulnerability to climate change. Diversity and Distributions 18(1):60-72. https://doi.org/10.1111/j.1472-4642.2011.00851.x

Dakos, V., and S. Kéfi. 2022. Ecological resilience: what to measure and how. Environmental Research Letters 17(4):43003. https://doi.org/10.1088/1748-9326/ac5767

Ewert, F., R. Baatz, and R. Finger. 2023. Agroecology for a sustainable agriculture and food system: from local solutions to large-scale adoption. Annual Review of Resource Economics 15(1):351-381. https://doi.org/10.1146/annurev-resource-102422-090105

Fernández-Habas, J., P. Sánchez-Zamora, F. Ceña-Delgado, and R. Gallardo-Cobos. 2018. Assessment of ecosystem services provision: the case of mountain olive groves in Los Pedroches, Southern Spain. New Medit 2:43-60. https://doi.org/10.30682/nm1802d

Ferreira, R., L. de Souza Cabral, R. D. Lins, G. Pereira e Silva, F. Freitas, G. D. Cavalcanti, R. Lima, S. J. Simske, and L. Favaro. 2013. Assessing sentence scoring techniques for extractive text summarization. Expert Systems with Applications 40(14):5755-5764. https://doi.org/10.1016/j.eswa.2013.04.023

Folke, C., S. R. Carpenter, B. Walker, M. Scheffer, T. Chapin, and J. Rockström. 2010. Resilience thinking: integrating resilience, adaptability and transformability. Ecology and Society 15(4):20. https://doi.org/10.5751/ES-03610-150420

Food and Agriculture Organization (FAO). 2018. The 10 elements of agroecology: guiding the transition to sustainable food and agricultural systems. FAO, Rome, Italy.

Food and Agriculture Organization (FAO). 2021. Making agrifood systems more resilient to shocks and stresses. FAO, Rome, Italy.

Food and Agriculture Organization (FAO). 2023. Urbanization, agrifood systems transformation and healthy diets across the rural-urban continuum. FAO, Rome, Italy.

Franco-Gaviria, F., M. Amador-Jiménez, N. Millner, C. Durden, and D. H. Urrego. 2022. Quantifying resilience of socio-ecological systems through dynamic Bayesian networks. Frontiers in Forests and Global Change 5:889274. https://doi.org/10.3389/ffgc.2022.889274

García-Vega, D., P. Dumas, R. Prudhomme, C. Kremen, and P.-M. Aubert. 2024. A safe agricultural space for biodiversity. Frontiers in Sustainable Food Systems 8:1328800. https://doi.org/10.3389/fsufs.2024.1328800

Gardner, E., T. D. Breeze, Y. Clough, H. G. Smith, K. C. R. Baldock, A. Campbell, M. P. D. Garratt, M. A. K. Gillespie, W. E. Kunin, M. McKerchar, S. G. Potts, D. Senapathi, G. N. Stone, F. Wäckers, D. B. Westbury, A. Wilby, and T. H. Oliver. 2021. Field boundary features can stabilise bee populations and the pollination of mass‐flowering crops in rotational systems. Journal of Applied Ecology 58(10):2287-2304. https://doi.org/10.1111/1365-2664.13948

Gliessman, S. 2016. Transforming food systems with agroecology. Agroecology and Sustainable Food Systems 40(3):187-189. https://doi.org/10.1080/21683565.2015.1130765

Hashemi, S.-Z., A. Darzi-Naftchali, F. Karandish, H. Ritzema, and K. Solaimani. 2022. Assessing agro-environmental sustainability of intensive agricultural systems. Science of the Total Environment 831:154994. https://doi.org/10.1016/j.scitotenv.2022.154994

Ingrisch, J., and M. Bahn. 2018. Towards a comparable quantification of resilience. Trends in Ecology & Evolution 33(4):251-259. https://doi.org/10.1016/j.tree.2018.01.013

Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES). 2019. Summary for policymakers of the global assessment report on biodiversity and ecosystem services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. S. Díaz, J. Settele, E. S. Brondízio, H. T. Ngo, M. Guèze, J. Agard, A. Arneth, P. Balvanera, K. A. Brauman, S. H. M. Butchart, K. M. A. Chan, L. A. Garibaldi, K. Ichii, J. Liu, S. M. Subramanian, G. F. Midgley, P. Miloslavich, Z. Molnár, D. Obura, A. Pfaff, S. Polasky, A. Purvis, J. Razzaque, B. Reyers, R. Roy Chowdhury, Y. J. Shin, I. J. Visseren-Hamakers, K. J. Willis, and C. N. Zayas, editors. IPBES, Bonn, Germany. https://files.ipbes.net/ipbes-web-prod-public-files/inline/files/ipbes_global_assessment_report_summary_for_policymakers.pdf

Kéfi, S., V. Guttal, W. A. Brock, S. R. Carpenter, A. M. Ellison, V. N. Livina, D. A. Seekell, M. Scheffer, E. H. van Nes, and V. Dakos. 2014. Early warning signals of ecological transitions: methods for spatial patterns. PLoS ONE 9(3):e92097. https://doi.org/10.1371/journal.pone.0092097

Kozar, R., R. Djalante, B. Leimona, S. M. Subramanian, and O. Saito. 2023. The politics of adaptiveness in agroecosystems and its role in transformations to sustainable food systems. Earth System Governance 15:100164. https://doi.org/10.1016/j.esg.2023.100164

Lang, J., S. Ponte, and T. Vilakazi. 2023. Linking power and inequality in global value chains. Global Networks 23(4):755-771. https://doi.org/10.1111/glob.12411

Lê, S., J. Josse, and F. Husson. 2008. FactoMineR an R Package for multivariate analysis. Journal of Statistical Software 25(1):1-18. https://doi.org/10.18637/jss.v025.i01

Lee, R. H., L. A. Navarro-Navarro, A. L. Ley, K. Hartfield, D. R. Tolleson, and C. A. Scott. 2021. Spatio-temporal dynamics of climate change, land degradation, and water insecurity in an arid rangeland: the Río San Miguel watershed, Sonora, Mexico. Journal of Arid Environments 193:104539. https://doi.org/10.1016/j.jaridenv.2021.104539

Levers, C., D. Müller, K. Erb, H. Haberl, M. R. Jepsen, M. J. Metzger, P. Meyfroidt, T. Plieninger, C. Plutzar, J. Stürck, P. H. Verburg, P. J. Verkerk, and T. Kuemmerle. 2018. Archetypical patterns and trajectories of land systems in Europe. Regional Environmental Change 18(3):715-732. https://doi.org/10.1007/s10113-015-0907-x

Lowder, S. K., M. V. Sánchez, and R. Bertini. 2021. Which farms feed the world and has farmland become more concentrated? World Development 142:105455. https://doi.org/10.1016/j.worlddev.2021.105455

Martin, E. A., B. Feit, F. Requier, H. Friberg, and M. Jonsson. 2019. Assessing the resilience of biodiversity-driven functions in agroecosystems under environmental change. Advances in Ecological Research 60:59-123. https://doi.org/10.1016/bs.aecr.2019.02.003

Marull, J., C. Cattaneo, S. Gingrich, M. G. de Molina, G. I. Guzmán, A. Watson, J. MacFadyen, M. Pons, and E. Tello. 2019. Comparative energy-landscape integrated analysis (ELIA) of past and present agroecosystems in North America and Europe from the 1830s to the 2010s. Agricultural Systems 175:46-57. https://doi.org/10.1016/j.agsy.2019.05.011

Meuwissen, M. P., P. H. Feindt, P. Midmore, E. Wauters, R. Finger, F. Appel, A. Spiegel, E. Mathijs, K. J. Termeer, A. Balmann, Y. Mey, and P. Reidsma. 2020. The struggle of farming systems in Europe: looking for explanations through the lens of resilience. EuroChoices 19(2):4-11. https://doi.org/10.1111/1746-692X.12278

Mijatović, D., F. van Oudenhoven, P. Eyzaguirre, and T. Hodgkin. 2013. The role of agricultural biodiversity in strengthening resilience to climate change: towards an analytical framework. International Journal of Agricultural Sustainability 11(2):95-107. https://doi.org/10.1080/14735903.2012.691221

Milazzo, F., P. Fernández, A. Peña, and T. Vanwalleghem. 2022. The resilience of soil erosion rates under historical land use change in agroecosystems of Southern Spain. Science of the Total Environment 822:153672. https://doi.org/10.1016/j.scitotenv.2022.153672

Mottet, A., A. Bicksler, D. Lucantoni, F. de Rosa, B. Scherf, E. Scopel, S. López-Ridaura, B. Gemmil-Herren, R. Bezner Kerr, J.-M. Sourisseau, P. Petersen, J.-L. Chotte, A. Loconto, and P. Tittonell. 2020. Assessing transitions to sustainable agricultural and food systems: a tool for agroecology performance evaluation (TAPE). Frontiers in Sustainable Food Systems 4:579154. https://doi.org/10.3389/fsufs.2020.579154

Niggli, U., M. Sonnevelt, and S. Kummer. 2023. Pathways to advance agroecology for a successful transformation to sustainable food systems. Pages 341-359 in J. von Braun, K. Afsana, L. O. Fresco, and M. Hassan, editors. Science and innovations for food systems transformation. Springer, Cham, Switzerland. https://doi.org/10.1007/978-3-031-15703-5_18

Oliver, T. H., E. Boyd, K. Balcombe, T. G. Benton, J. M. Bullock, D. Donovan, G. Feola, M. Heard, G. M. Mace, S. R. Mortimer, R. J. Nunes, R. F. Pywell, and D. Zaum. 2018. Overcoming undesirable resilience in the global food system. Global Sustainability 1:e9. https://doi.org/10.1017/sus.2018.9

Oliver, T. H., M. S. Heard, N. J. B. Isaac, D. B. Roy, D. Procter, F. Eigenbrod, R. Freckleton, A. Hector, C. D. L. Orme, O. L. Petchey, V. Proença, D. Raffaelli, K. B. Suttle, G. M. Mace, B. Martín-López, B. A. Woodcock, and J. M. Bullock. 2015. Biodiversity and resilience of ecosystem functions. Trends in Ecology & Evolution 30(11):673-684. https://doi.org/10.1016/j.tree.2015.08.009

Page, M. J., J. E. McKenzie, P. M. Bossuyt, I. Boutron, T. C. Hoffmann, C. D. Mulrow, L. Shamseer, J. M. Tetzlaff, E. A. Akl, S. E. Brennan, R. Chou, J. Glanville, J. M. Grimshaw, A. Hróbjartsson, M. M. Lalu, T. Li, E. W. Loder, E. Mayo-Wilson, S. McDonald, L. A. McGuinness, L. A. Stewart, J. Thomas, A. C. Tricco, V. A. Welch, P. Whiting, and D. Moher. 2021. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 372:n71. https://doi.org/10.1136/bmj.n71

Peterson, C. A., V. T. Eviner, and A. C. Gaudin. 2018. Ways forward for resilience research in agroecosystems. Agricultural Systems 162:19-27. https://doi.org/10.1016/j.agsy.2018.01.011

Quinlan, A. E., M. Berbés-Blázquez, L. J. Haider, and G. D. Peterson. 2016. Measuring and assessing resilience: broadening understanding through multiple disciplinary perspectives. Journal of Applied Ecology 53(3):677-687. https://doi.org/10.1111/1365-2664.12550

Ruiz-Agudelo, C. A., O. D. Bonilla-Uribe, and C. A. Páez. 2015. The vulnerability of agricultural and livestock systems to climate variability: using dynamic system models in the Rancheria upper basin (Sierra Nevada de Santa Marta). Journal on Protected Mountain Areas Research and Management 7(1):50-60. http://epub.oeaw.ac.at/eco.mont https://doi.org/10.1553/eco.mont-7-2s50

Schiavo, M., C. Le Mouël, X. Poux, and P.-M. Aubert. 2023. The land use, trade, and global food security impacts of an agroecological transition in the EU. Frontiers in Sustainable Food Systems 7:1189952. https://doi.org/10.3389/fsufs.2023.1189952

Schlüter, M., G. Caniglia, K. Orach, Ö. Bodin, N. Magliocca, P. Meyfroidt, and B. Reyers. 2022. Why care about theories? Innovative ways of theorizing in sustainability science. Current Opinion in Environmental Sustainability 54:101154. https://doi.org/10.1016/j.cosust.2022.101154

Schlüter, M., B. Müller, and K. Frank. 2019. The potential of models and modeling for social-ecological systems research: the reference frame ModSES. Ecology & Society 24(1):31. https://doi.org/10.5751/ES-10716-240131

Seppelt, R., C. Arndt, M. Beckmann, E. A. Martin, and T. W. Hertel. 2020. Deciphering the Biodiversity-Production mutualism in the global food security debate. Trends in Ecology & Evolution 35(11):1011-1020. https://doi.org/10.1016/j.tree.2020.06.012

Serfilippi, E., and G. Ramnath. 2018. Resilience measurements and conceptual frameworks: a review of the literature. Annals of Public and Cooperative Economics 89(4):645-664. https://doi.org/10.1111/apce.12202

Standish, R. J., R. J. Hobbs, M. M. Mayfield, B. T. Bestelmeyer, K. N. Suding, L. L. Battaglia, V. Eviner, C. V. Hawkes, V. M. Temperton, V. A. Cramer, J. A. Harris, J. L. Funk, and P. A. Thomas. 2014. Resilience in ecology: abstraction, distraction, or where the action is? Biological Conservation 177:43-51. https://doi.org/10.1016/j.biocon.2014.06.008

Suškevičs, M., K. Karner, C. Bethwell, F. Danzinger, S. Kay, T. Nishizawa, J. Schuler, K. Sepp, R. Värnik, M. Glemnitz, M. Semm, C. Umstätter, T. Conradt, F. Herzog, N. Klein, T. Wrbka, P. Zander, and M. Schönhart. 2023. Stakeholder perceptions of agricultural landscape services, biodiversity, and drivers of change in four European case studies. Ecosystem Services 64:101563. https://doi.org/10.1016/j.ecoser.2023.101563

Tendall, D. M., J. Joerin, B. Kopainsky, P. Edwards, A. Shreck, Q. B. Le, P. Kruetli, M. Grant, and J. Six. 2015. Food system resilience: defining the concept. Global Food Security 6:17-23. https://doi.org/10.1016/j.gfs.2015.08.001

Ticktin, T., S. Quazi, R. Dacks, M. Tora, A. McGuigan, Z. Hastings, and A. Naikatini. 2018. Linkages between measures of biodiversity and community resilience in Pacific Island agroforests. Conservation Biology 32(5):1085-1095. https://doi.org/10.1111/cobi.13152

Tittonell, P. 2020. Assessing resilience and adaptability in agroecological transitions. Agricultural Systems 184:102862. https://doi.org/10.1016/j.agsy.2020.102862

Tittonell, P. A., J. de Grazia, S. de Hek, and E. Bricchi. 2006. Exploring land use scenarios by long-term simulation of soil organic matter in central Argentina. Spanish Journal of Agricultural Research 4(4):381-389. https://doi.org/10.5424/sjar/2006044-215

United Nation Statistic Division. 2024. Methodology: standard country or area codes for statistical use (M49): geographical regions. UN Statistic Division, New York, New York, USA. https://unstats.un.org/unsd/methodology/m49/

United Nation University-Institute for the Advanced Study of Sustainability (UNU-IAS). 2024. Indicators of resilience in socio-ecological production landscapes and seascapes (SEPLS): 2024 Edition. UNU-IAS, Tokyo, Japan.

van der Lee, J., D. Kangogo, Ş. Ö. Gülzari, D. Dentoni, S. Oosting, J. Bijman, and L. Klerkx. 2022. Theoretical positions and approaches to resilience assessment in farming systems. A review. Agronomy for Sustainable Development 42:27. https://doi.org/10.1007/s13593-022-00755-x

van Oudenhoven, F. J., D. Mijatović, and P. B. Eyzaguirre. 2011. Social‐ecological indicators of resilience in agrarian and natural landscapes. Management of Environmental Quality: An International Journal 22(2):154-173. https://doi.org/10.1108/14777831111113356

Vanwindekens, F. M., A. Gobin, Y. Curnel, and V. Planchon. 2018. New approach for mapping the vulnerability of agroecosystems based on expert knowledge. Mathematical Geosciences 50(6):679-696. https://doi.org/10.1007/s11004-018-9730-5

Vialatte, A., C. Barnaud, J. Blanco, A. Ouin, J.-P. Choisis, E. Andrieu, D. Sheeren, S. Ladet, M. Deconchat, F. Clément, D. Esquerré, and C. Sirami. 2019. A conceptual framework for the governance of multiple ecosystem services in agricultural landscapes. Landscape Ecology 34(7):1653-1673. https://doi.org/10.1007/s10980-019-00829-4

Walker, B., C. S. Holling, S. R. Carpenter, and A. P. Kinzig. 2004. Resilience, adaptability and transformability in social-ecological systems. Ecology and Society 9(2):5. https://doi.org/10.5751/ES-00650-090205

Wezel, A., B. G. Herren, R. B. Kerr, E. Barrios, A. L. R. Gonçalves, and F. Sinclair. 2020. Agroecological principles and elements and their implications for transitioning to sustainable food systems. A review. Agronomy for Sustainable Development 40:40. https://doi.org/10.1007/s13593-020-00646-z