Avenues of archetype analysis: roots, achievements, and next steps in sustainability research

Recent years have seen a proliferation of studies that use archetype analysis to better understand and to foster transitions toward sustainability. This growing literature reveals a common methodological ground, as well as a variety of perspectives and practices. In this paper, we provide an historical overview of the roots of archetype analysis from ancient philosophy to recent sustainability science. We thereby derive core features of the archetype approach, which we frame by eight propositions. We then introduce the Special Feature, “Archetype Analysis in Sustainability Research,” which offers a consolidated understanding of the approach, a portfolio of methods, and quality criteria, as well as cutting-edge applications. By reflecting on the Special Feature’s empirical and methodological contributions, we hope that the showcased advances, exemplary applications, and conceptual clarifications will help to design future research that contributes to collaborative learning on archetypical patterns leading toward sustainability. The paper concludes with an outlook highlighting central directions for the next wave of archetype analyses.


INTRODUCTION
When searching for ways to achieve sustainability, scholars are confronted by the diversity of local and regional factors and processes, which are often interlocked or telecoupled in complicated ways. How can sustainability problems and solutions, in all their diversity, be explained? To address such challenges for research and practice, recent years have seen the rise of archetype analysis as a novel approach in sustainability research .
If sustainability research does not take care of each case's specific features, we might continue on the track of blueprint solutions. Such panaceas are problematic if they disregard socio-cultural, political-institutional, economic, and ecological particularities of the cases and contexts in which they guide action ). On the other hand, there are common patterns around the world that repeatedly shape the (un)sustainability of socialecological systems. For example, Sietz et al. (2011) take the zaï technique, a traditional land rehabilitation approach used in Burkina Faso. This is a region-specific solution that has helped to improve degraded soils, food production, and well-being and to reduce rural out-migration. Yet, this raises the question as to whether the zaï technique is replicable to restore farmlands elsewhere. This highlights the importance of identifying the relevant social-ecological similarities among various locations to assess such a transfer of solutions.
To address questions of similarity and replicability in a rigorous way, archetype analysis can draw from a broad portfolio of methods . Knowledge of archetypical patterns across cases has supported a better understanding of key sustainability challenges related to land use, climate change adaptation, vulnerability, large-scale land acquisition, ecological footprints, and regional development, among others. Archetypes featured prominently in the UNEP's (2007) Global Environment Outlook 4. Currently, the development of best practices for archetype analysis is one frontier of innovation . On the following pages, we provide a novel consolidated understanding of the approach, including a sketch of its emergence and core ideas. Then we introduce the Special Feature, "Archetype Analysis in Sustainability Research" of Ecology and Society. This Special Feature contains several conceptual and methodological papers, and it showcases cutting-edge examples of archetype analyses. We hope that this collection will help readers to judge the promises and pitfalls of the archetype approach.
"Archetype analysis investigates recurrent patterns of the phenomenon of interest at an intermediate level of abstraction to identify multiple models that explain the phenomenon under particular conditions" . A core starting point is the assumption that it is useful to consider multiple explanations, theories, or models in parallel. An archetype analysis thus determines or studies a suite of (i.e., multiple) patterns, each called an archetype. A single archetype would not be able to explain all instances of a phenomenon (Eisenack et al. 2006a). If multiple archetypes on the same issue are not admitted, i.e., if research aims at a general law that holds for the complete universe of cases, research risks a meaningless overgeneralization. Furthermore, not admitting a suite of multiple archetypes might hamper interdisciplinary collaboration when a common epistemological or ontological base is not available (Spash 2012). We do not consider archetype analysis as a method because it is open to many empirical and analytical techniques. The term "approach" seems more appropriate because patterns can be analyzed using different qualitative, quantitative, and mixed methods . Implicit in this understanding is a focus on cases as units of analysis. What makes a case depends on the phenomenon of interest or research question: it may be a region, a grid cell, an organization, a conflict, a household, a land or food system, or a kind of contract, for example. Whereas some archetype analyses deal with a large number of cases, others provide rich detail on a small number. Archetype analysis is a comparative approach to deal with a medium or large number of heterogeneous cases. 2 An archetype analysis produces a suite of archetypes (and not a universal law). 3 An archetype characterizes components of cases (e.g., actors, processes, effects, or sub-systems), that re-occur in multiple (but not necessarily all) cases. 4 Archetypes function as building blocks that can be combined in different ways to explain individual cases. 5 A comprehensive archetype analysis characterizes each archetype by three elements: (i) a configuration of attributes; (ii) a theory or hypothesis that explain the relation between the attributes; (iii) a set of cases where it holds. 6 All archetypes in a suite characterize cases and configurations by referring to a common vocabulary of attributes (but each archetype does not need to use it completely). 7 Attributes and archetypes are formulated on an intermediate level of abstraction. 8 Archetypes are analytical or mental constructs, but are not necessarily material or functional mechanisms or systems.

Early roots
Aristotle (fourth century BC) was likely not the first to sort particular cases into (more abstract) different kinds, types, or classes, where each class has something in common. Archetype analysis classifies cases by presuming that there is likely no single mechanism behind all of them (prop. 2). For Aristotle, "kinds" are characterized by properties that are essential to all contained individuals, while other properties are accidental (Aristotle, Metaphysics). Classification only makes sense if there are multiple classes (prop. 2, 3). Things can be organized in a hierarchy of concepts with different degrees of generality or abstraction (Aristotle, Organon; cf. prop. 7). When scholars address a (new) set of phenomena, it has always been common to first develop classifications, for instance the Linneean taxonomy of living beings, the periodic table of chemical elements, or the classification of stars by the astronomer Angelo Secchi. The semantic of classes and concepts can be made precise by distinguishing the extension and intension of a concept (Carnap 1956). Whereas the former is the set of all particular "objects" (members) of the concept (e.g., all cases sharing a particular attribute), the latter is the set of properties or attributes that all objects belonging to the concept share. In archetype analyses, when cases (extensions) are described, we call the properties used (intension) the "attributes" (prop. 6), and aim to find recurring configurations of those attributes (prop. 5). Depending on whether the attributes' extension is larger or smaller, archetypes can differ in how general or abstract they are (prop. 7).
John Locke (since 1689) was, to our knowledge, the first to work with the term "archetype," as "patterns or models from which they [real ideas] are copied" (Locke 1689, An Essay Concerning Human Understanding). Because classification is essential for archetype analysis, an obvious question follows: What is different between an archetype and a type? For Locke, "archetypes are made by the mind to serve as standards for classifying and naming things" (prop. 8). Such classes are required if humans aim at knowing an overly complicated world of particulars. Classes do not serve this purpose if they contain just single particulars (prop. 3). Because archetypes are made by the mind, Locke rejects essential properties. George Berkeley (1710, A Treatise Concerning the Principles of Human Knowledge) even understands them as purely subjective. For Locke, archetypes are not arbitrary because they need to "conform to the real being and existence of things". Although there are many ways to classify one same subject matter (prop. 2), Locke's conception of archetypes can be understood as constrained and convergent (Anstey 2011); they are thus more than arbitrary types, but need to fit to theoretical and empirical constraints. Archetypes, as we understand them, might be characterized as "cluster kinds" (Boyd 1991): configurations of attributes that are held together by some internal or external mechanisms. They are thus not arbitrary and each archetype can be underpinned by a theory to explain such a mechanism (prop. 5).
Immanuel Kant (since 1781) uses the term "intellectus archetypus," yet in another way than we do, to denote the idea of a divine being or perfectly rational mode of reasoning (Kant 1781, Critique of Pure Reason). In English translations, the term "archetypal ideas" is also used when Kant discusses biological kinds (Kant 1790, Critique of Judgement). Each individual organism from a species or genus might be conceived as being a copy of a common Urbild (archetype). Kant understands these archetypes only as a heuristic that is helpful for reasoning.
Carl Jung (since 1902) developed well-known psychological archetypes of personal traits. They refer to common symbolic patterns that are inscribed into the collective unconscious. This https://www.ecologyandsociety.org/vol26/iss2/art31/ is a more domain-specific understanding in psychology than the one used in sustainability research. Furthermore, archetype analysis in sustainability research does not aim to reveal universal patterns (prop. 1).
Middle-range theories (since 1968) were proposed for the social sciences (Merton 1968) to contrast grand theories or general laws (prop. 2). Middle-range theories aim at explaining empirical regularities, but within a clearly delimited range of conditions ). Yet, such theorizing "always involves more than the (detailed) description of single cases" (Esser 2010:1;i.e., prop. 3). This can be done by hypothesizing mechanisms or causal relations (prop. 5). The mechanisms should play a role for multiple cases. Although some scholars envisaged a research program to come up, over time, with a coherent collection of middle-range theories to cover large parts of social inquiry, this vision has not gained traction (Esser 2002). Yet, middle-range theories are developed for specific research questions or objectives (prop. 1). Furthermore, Esser (2002) demonstrates how different mechanisms from social theory (see also Elster 2006) can be combined in a modular way (prop. 4), tailored to obtain middlerange theories for specific issues. Archetype analysis is one approach to develop middle-range theories (Magliocca et al. 2018. Vice versa, middle-range theories can offer explanations that characterize archetypes (prop. 5).
The Pattern Language (Alexander et al. 1977) for architecture is an inspiring approach related to our conceptualization of archetypes. Taking a design perspective, it does not provide a universal theory of architecture, but provides a large collection of architectural forms and patterns that frequently recur in buildings (e.g., "cascade of roofs," "roof garden," "six-foot balcony"). Each pattern is supplied with an empirical or theoretical justification (prop. 5). For new architectural projects, the authors suggest to choose some (but not all) of the patterns, like choosing words from a language when you compose a text (prop. 6). The patterns serve as building-blocks, because they can be combined in different ways for idiosyncratic projects (prop. 4).
Other related streams of thought, which are not covered by our above propositions, may warrant more study in the future. The ordinary language understanding of "pattern" might well be a good starting point (e.g., Kelso 1997), because it also admits classification with non-crisp or fuzzy boundaries. We might say that two cases belong to the same pattern if they are not similar in kind, but in degree. Also, ideal types of Weber (1922) admit comparison by degree. Recurrence of archetypes might also be understood by family resemblance (Wittgenstein 1953).

Related branches in sustainability research
Since the 1980s, some approaches similar to archetype analysis have been suggested, partly being influential in sustainability research and in the development of archetype analysis. Configurational approaches (since 1987) aim at identifying attributes that tend to co-occur in the case universe, but not necessarily in all cases (prop. 3). Ragin's (1987) comparative method, for instance, takes account of idiosyncratic properties of cases, but also identifies (sets of) conditions that are sufficient for the outcome of interest in one or more cases (prop. 3, prop. 7). The method is strong in the study of a medium number of cases where inferential statistics are not appropriate (prop. 1). Although each case is studied in depth (also to provide explanations for such conditions, prop. 5), all cases are also more abstractly described by a common set of conditions (yet, prop. 6 does not hold, because every case needs to be described by the complete set of conditions). Qualitative comparative analysis (QCA; see also Schneider and Wagemann 2012), probably the most prominent configurational approach, derives a general Boolean formula from the data. QCA outputs expressions in disjunctive normal form, i.e., like "the outcome holds in all cases where A OR B OR C...", where a list of alternative (sets of) conditions (A, B, C, ...) is linked by the Boolean "OR" operator. This is called equifinality, i.e., that different (sets of) conditions can lead to the same outcome (prop. 2).
System archetypes (since 1990) have been developed in the domain of systems dynamics (Senge 1990, Wolstenholme 2003. Each archetype (prop. 2) describes a common management problem, and is formulated by using a causal-loop diagram (prop. 5).
Examples are the "Tragedy of the Commons" or "Limits to Growth". The variables in the diagrams need to be quite abstract in order to be applicable in many situations (e.g., "Resources," "Effort," "Capacity"), leading to interesting discussions on conceiving generic structure (Lane and Smart 1996). This requires the analyst to give variables a more concrete meaning, e.g., an indicator, when applying a system archetype (prop. 7). Causal loops characterize mechanisms that determine how variables change over time, a possibility not used by many archetype analyses so far. System archetypes further assume that different real-world problems, which are appropriately captured by the same archetype, can be addressed by similar generic solutions, but that not all real-world problems are of the same kind (prop. 3).
Syndromes of global change (since 1994), inspired by the medical metaphor, were proposed by the German Advisory Council on Global Change (WBGU 1994) to map configurations of cooccurring factors (symptoms) that constitute syndromes. Syndromes decompose global change dynamics into coevolutionary trends that appear repeatedly in typical combinations (Lüdeke et al. 2004). A suite of 16 syndromes (prop. 2) were identified, including "Urban Sprawl," "Sahel Syndrome," or "Overexploitation" (cf. Kropp et al. 2006, Sietz et al. 2006, Reckien et al. 2011) that reappear in many regions (prop. 3). Drawing on the same list of about 80 symptoms (prop. 6), each syndrome is characterized by a set of closely related symptoms (prop. 5). The dynamics of some syndromes have been analyzed with qualitative differential equations (e.g., Petschel-Held and Lüdeke 2001, Eisenack et al. 2006b, Sietz et al. 2006, Reckien et al. 2011 were also inspired by the medical metaphor (Young 2002, Young et al. 2006. The approaches highlight that sustainability problems are too diverse (prop. 1) to be addressed by simple institutional "panaceas" (like private property, Ostrom 2007. They aim at "identifying key features of needs for governance on a case-by-case basis ... in order to ensure a good match ... for ... the institutional response" (Young 2019). Yet, institutional design principles might be transferred between cases if they are diagnosed in the same way (prop. 2, 3; Cox et al. 2010). This requires partitioning the universe of cases (prop. 2). The Social-Ecological Systems framework (SES; Ostrom 2009, McGinnis andOstrom 2014) brings different levels of abstraction (prop. 7) more to the center. The framework proposes a nested list of variables that are considered to be commonly important for many social-ecological-technical systems (a common vocabulary, prop. 6). Each variable in the most abstract "tier" (e. g., "resource units," "governance system") is refined to a more concrete set of about 3-9 variables (e.g., "number of resource units," "monitoring and sanctioning rules"). Not all variables are relevant for a particular case, and they can be further refined to higher tier variables if required. First archetype studies build on this set of variables ( 3), e.g., related to global commons, contaminated sites, energy production, or urbanization in coastal zones. The archetypes confirm essential mechanisms described by the syndromes of global change. For instance, the Dryland Archetype includes the poverty-degradation spiral, the typical mechanism of the Sahel Syndrome (Petschel-Held et al. 1999). Methodologically, basic vulnerabilitycreating conditions were synthesized from the literature and hypotheses about mechanisms were deduced (prop. 5). Causal variables were expressed by quantitative indicators and clustered (prop. 5) in mutually exclusive vulnerability profiles (Sietz et al. 2011, Kok et al. 2016). The vulnerability profiles were further differentiated at local scales (Sietz et al. 2012, Vidal Merino et al. 2019 and nested continental scales (Sietz et al. 2017); they thus navigate different scales of abstraction (prop. 7). Policy implications and entry-points for improvement were discussed for each vulnerability profile, and the transferability of findings was demonstrated in combination with ground-truthing (testing and contextualizing) against local case studies.
After having compared these influential approaches in light of the core propositions, we now relate this conceptual consolidation of archetype analysis to the genesis and main insights of this Special Feature.

SPECIAL FEATURE ON ARCHETYPE ANALYSIS IN SUSTAINABILITY RESEARCH
The rapid growth and diversification of studies using archetype analysis has generated variations and inconsistencies about using the approach . This Special Feature on "Archetype Analysis in Sustainability Research" was thus created to confront several major research challenges, including a lack of consensus on the multiple and precise meanings of the approach in sustainability research, on the portfolio of methods, and on best practice examples for high-quality archetype analysis. It thus aims to, Overall, 16 papers from international author teams appear in this Special Feature, which cover different sustainability issues (see Table 2 for a summary). Three opening synthesis papers provide an overview from different angles.  provide a consolidated perspective on the multiple meanings, motivations, and policy relevance of archetype analysis, based on the literature so far. Eisenack et al. (2019) analyze main challenges for archetype analyses in order to develop quality criteria and to improve future archetype analyses. Sietz et al. (2019) synthesize the portfolio of methods currently used for the approach, and highlight the strengths, weaknesses, and analytical frontiers in employing them Ecology and Society 26(2): 31 https://www.ecologyandsociety.org/vol26/iss2/art31/ The more substantive and topic-related papers that follow the synthesis papers use a range of practices and approaches. It appears that any standards that may emerge from the papers depend on the kind of objectives, or possible style, of archetype analysis, which warrant different methodological decisions. The papers exemplify different kinds of objectives. Newig et al. (2019) use archetype analysis to develop a theoretical argument about sustainability pathways. Several studies work with original empirical material to identify a whole suite of archetypes to explain or sort out phenomena in their cases (Karrasch et al. 2019, Moser et al. 2019, Villamayor-Tomas et al. 2020b, Wang et al. 2019. Other papers identify whole suites of archetypes through meta-studies of secondary sources (Gotgelf et al. 2020, Horcea-Milcu et al. 2020, Tribaldos et al. 2020. A further set of papers use archetypes in a deductive way to design or analyze scenarios , Sitas et al. 2019. Neudert et al. (2019) start from a suite of archetypes identified in earlier studies in order to refine them, to diagnose patterns in a new set of cases, and to dig deeper into selected archetypes.
Overall, it appears that archetype analyses can function to identify and explain patterns, or to diagnose new cases, or to develop scenarios .
In addition to methodological advances, the contributions advance understandings of sustainability issues. They range from land use, climate change adaptation, ecosystems and biodiversity, institutional failure, and socioeconomic pathways to studies of the science-policy interface (Table 2). Although each paper engages deeply with its specific field and provides multiple specific insights, their collection in the present Special Feature offers a number of overarching insights into sustainability.
First, the papers offer new insights into pathways of adaptation and transformations. Newig et al. (2019) show how institutional decline and failure can act as drivers of sustainability transformations in five surprising ways. Wang et al. (2019) demonstrate how institutional fit and decentralized governance can facilitate rural renewal in China. Villamayor-Tomas et al. (2020b) find that institutions are more relevant for adapting to drought situations when they specifically address water scarcity, compared to more generic institutions. Different institutional regime types are linked with different adaptation paths. Gotgelf et al. (2020) show that the generation of climate information through science-policy-practice partnerships may trigger transformational adaptation, e.g., through archetypical processes of learning about long-term implications of climate change.
Second, archetypes can function as boundary objects in analyses that coherently use multiple methods to examine archetypical social-ecological interactions operating at multiple scales. For example, Magliocca et al. (2019) provide a proof-of-concept analyzing the local causes and consequences of land acquisitions in Cambodia.
Third, archetypes help in formulating policy recommendations that are tailored to specific diagnoses, e.g., local financial constraints in climate change adaptation (Moser et al. 2019). Neudert et al. (2019) show that interactions among archetypes can be crucial for deriving more refined recommendations compared to those derived from considering archetypes separately.
Fourth, scenario archetypes have supported science-policy dialogues such as in the Intergovernmental Science-Policy Platform for Biodiversity and Ecosystem Services in multiple fruitful ways. The studies suggest that a next generation of scenario archetypes might be more inductive and consider linkages of scenarios across scales in a more detailed way , Sitas et al. 2019.
Finally, archetypes are useful to rethink positions in sustainability research. Karrasch et al. (2019) assess stakeholder perceptions of land-use and ecosystem services to identify conflicting perspectives in participatory processes. Horcea-Milcu et al. (2020) identify points of convergence and divergence of archetypal concepts in research on sustainability transformations, and Tribaldos et al. (2020) show under which conditions intense stakeholder interactions in knowledge coproduction in different phases of research is associated with sustainable impacts.

CONCLUSIONS AND OUTLOOK TO A NEW GENERATION OF ARCHETYPE ANALYSIS
Archetype analysis is becoming an increasingly prominent approach in sustainability research. This Special Feature showcases a consolidated perspective on the multiple meanings, methods, and quality criteria of this approach and a variety of cutting-edge applications. At the same time, the collection of papers demonstrates several promising avenues for future archetype analysis. In closing, we wish to highlight three of them.
First, we have not reached established methods for conducting dynamic archetype analysis. Most of the current studies do not trace patterns over time. Yet, speaking of causality, or of a transition toward sustainability, entails considering patterns of change. This requires a clear conceptual understanding, e.g., whether to consider sequences of (static) archetypes, or archetypical sequences of system states. Common vocabularies for classification of case-specific dynamic mechanisms likely need to distinguish different time scales. Dynamic archetype analysis could also benefit from a methodological portfolio that builds on tools from scenario analysis (e.g., Rhyne 1995, Weimer-Jehle 2006, Schmid et al. 2017, Kemp-Benedict et al. 2019, Kearney 2021, or can trace patterns over time on a medium level of abstraction (e.g., Kuipers 1994, Wolstenholme 2003, Eisenack et al. 2006a, 2007, Sietz et al. 2006, Zitek et al. 2009, Reckien et al. 2011, Sietz 2014).
Second, we still need advances in transferring or scaling sustainability solutions with archetype analysis. For evidencebased insights into social-ecological similarities, e.g., if archetypes are being shared across cases, it is important to consider both geographical and social scales to ensure that solutions fit the various stakeholders' demands, preferences, and livelihood strategies, as, e.g., for replicating the zaï technique. The concept of scaling domains, i.e., regions with similar social-ecological potentials and constraints where solutions can be tested (Coe et al. 2014), might support systematic procedures for the out-scaling, transfer, replication, or adoption of successful solutions by an increasing number and range of stakeholders who operate under https://www.ecologyandsociety.org/vol26/iss2/art31/ similar social-ecological conditions (Douthwaite et al. 2007).
With the concept of generic leverage points for interventions (Chan et al. 2020), future archetype analysis may also spur the up-scaling (Linn 2012) of sustainability solutions or anchoring them at higher institutional levels responsible for promoting change.
Third, sustainability research currently sees many archetype studies that investigate particular phenomena separately. It is also possible that a sequence of archetype analyses discovers more details about shared phenomena of interest, and expands the set of cases where the identified archetypes hold, leading to cumulative learning. Currently, such learning is still limited, despite its recognized importance in sustainability research (Alexander et al. 2020, Newig and Rose 2020, Pauliuk 2020, Villamayor-Tomas et al. 2020a). Therefore, a next generation of archetype research may consolidate, validate, and refine knowledge about archetypes in specific domains, for instance sustainable land use, climate adaptation, and biodiversity. To this end, future studies can take up existing knowledge and unlock the potential of archetype analysis for cumulative learning about patterns of (un)sustainability.
Responses to this article can be read online at: https://www.ecologyandsociety.org/issues/responses. php/12484