Introduction

Effective policy-making around complex social–environmental challenges requires development of mid-range theories straddling generality, realism, and precision across diverse explanatory variables (Meyfroidt et al. 2018). Middle range theories are “contextual generalisations that describe chains of causal mechanisms explaining a well bounded range of phenomena, as well as conditions that enable, trigger, or prevent these causal chains” (Meyfroidt 2016). A diagnostic approach has been long considered effective in developing such contextual generalisations (Cox 2011). The Institutional Analysis and Development framework (IADF) (Ostrom 1990) and its ecologically grounded successor (Cole et al. 2019a; b), the Social Ecological Systems Framework (SESF) (McGinnis and Ostrom 2014; Ostrom 2009) are powerful tools in this context, at the core of which lies the concept of action situations (ASs). Articulated in later versions of the SESF (Ostrom and Cox 2010; Cole et al. 2019a, b), the AS is a complex of actor–resource interactions—influenced by four key components (or first-tier components of the SESF): Resource Systems (RSs), Resource Units (RUs), Governance Systems (GSs) and Actors (As). ASs represent the space where policy decisions are devised based upon the actor’s relative positions within the complex as well as the various rules that they are constrained or enabled by (McGinnis 2011). A focal AS represents patterns of interactions amongst actors and ecosystem resources within the system of interest. These interactions include social and ecological components, each of which can further be decomposed into smaller elements, as well as be situated within the context of broader aggregations (McGinnis 2011). Despite their utility, challenges in applying the IADF and SESF persist, particularly from the perspective of mid-large-scale social-ecological systems (SES) (Villamayor-Tomas et al. 2020; Epstein et al. 2020; Cole et al. 2019a, b; Partelow 2018; Thiel et al. 2015), due to a gap in developing consistent techniques to interpret and operationalize the large number of variables contained within them (Cox 2014; Leslie et al. 2015; Delgado-Serrano and Ramos 2015; Cumming et al. 2020).

SES challenges further consist of the need to acknowledge and address multiple, interdependent ASs where the outcome of one AS can influence trajectories or outcomes of other ASs. This phenomenon has been explained through the networks of adjacent action situations (NAASs) concept developed by McGinnis (2011). Expanding upon the concept of ASs, at the core of the NAASs lie interactions between the four first-tier components of the SESF described earlier. However, studies have pointed out (Cox 2008; Cox et al. 2020; Hinkel et al. 2015; Vogt et al. 2015; Epstein et al. 2013) that two of these components: the RS and RU remain insufficiently decomposed, challenging the utility of applying the NAAS concept to mid-large-scale SESs.

In this paper, we engage with these two gaps—lack of consistency in applying the SESF and the linked concern of insufficient decomposition of RS and RU. We do this by (a) introducing the concept of nested resource system (NRSs) to negotiate complexity of RS–RU interactions, (b) developing a diagnostic approach to applying the NRS within mid-large-scale SESs, and (c) identifying spatially situated NAASs operating within NRS. We focus explicitly and strategically on the RS and RU components of the SESF. We then provide a diagnostic tool that enables a standardised approach to describing and analysing SESs, both from the perspective of smaller, well-defined SESs as well as mid-large-scale NRS. We demonstrate the applicability of our diagnostic process through comparison across two diverse and distinct systems—networks of lakes in south Indian Bengaluru (Unnikrishnan et al. 2016, 2020) and German winter wheat breeding systems (Gerullis et al. 2021).

The context

Application of the SESF to cases requires a three-tiered process—(a) selecting the focal level of analysis; (b) selecting variables to be measured and the implementation of indicators for those variables (data collection and analysis), and (c) communicating results of the analysis across research communities through a common base of shared terms (McGinnis and Ostrom 2014). In mid-large-scale SESs, one often finds that it is difficult to both draw systemic boundaries as well as specify which components of the SES become the RS and RU and in what context. We argue that this challenge arises because mid-large-scale SESs are inherently nested wherein multiple SESs may exist within each other and are bounded by a larger SES, while not necessarily being linked or networked with each other. This observation was first articulated by Cox (2010, 2014) in his application of the SESF to the Taos acequia irrigation system.

As an example, if we consider the Yellowstone National Park as our system of interest (and therefore the RS), this does not automatically imply that other components of the park exist solely as RUs within that RS. Yellowstone National Park contains multiple potential SESs nested (but not necessarily networked with reference to how system boundaries are defined, or the question being investigated) within it—lakes, rivers, grasslands, calderas, that may or may not be hierarchical in their organisation with reference to the park. Therefore, there is a need to explicitly decompose the RS and RU into possible further subcomponents (Cox 2010, 2014). Multiple RUs and RSs may be involved in equally numerous ASs; further, diverse institutional arrangements may affect multiple ASs simultaneously (Villamayor-Tomas et al. 2015). NAASs operating in such SESs are thus usually scattered not just along societal and institutional dimensions, but also along spatial and ecological ones.

Several approaches to addressing these challenges have been proposed—for example, the addition of a seventh core subsystem category to the SESF—that of ecological rules, allowing analysts to incorporate ecologically derived knowledge into their cases (Epstein et al. 2013). Oberlack et al. (2018) advance the idea of telecoupled RS which refer to RS connections across multiple SESs spread across large distances. Cole et al. (2019a; b) defines processes by which social, ecological, and institutional factors interact across ASs producing social–ecological outcomes, through combining the SESF with the IADF (Cole et al. 2019a; b). Schlüter et al. (2019a; b) have extended the NAAS approach to include explicit consideration of relations that exist between human and non-human entities; in other words, between social and the ecological components of the SES. In doing so, they propose the Social Ecological Action Situation (SEAS) framework, which recognises three distinct forms of ASs, namely the Social AS, Ecological AS, and the Social–ecological AS (Schlüter et al. 2019a; b). Möck et al. (2019) propose that spatial scales, temporal change within systems, and resource linkages may be integrated through an approach of layering ASs as an analytical technique in applying the IADF (as opposed to the conventional technique of comparing temporally and spatially fixed aspects of the ASs).

We argue that while each of these approaches add a lot of value, they do not, however, engage with the root challenge of reconciling decomposition of the RS and RUs. The concept of telecoupled RSs (Oberlack et al. 2018) while coming close to this decomposition does not engage with the idea that multiple systems can exist embedded within each other but might not always be connected in their processes and functions—consider, for example, our earlier discussion of the Yellowstone National Park. This means that NAASs operating within these first-tier components remain spatially aggregated, implying that the links between RSs and their spatial dimensions still need to be explicitly recognized. Further, an analyst applying the SESF with the objective of comparing across diverse cases against a broader goal of generating middle range theories, would need a standardised approach towards unpacking and describing the complexity of their cases both from the perspective of a decomposed RS and RUs as well as the complex array of NAASs that emerge from these contexts.

To address these gaps, we first articulate in greater detail the idea of NRS. We then build upon and expand the diagnostic procedure developed by Hinkel et al. (2015) to include considerations of NRS. This distinction allows us to account for multiple, simultaneously occurring NAASs that collectively give rise to SES outcomes.

Nested resource systems (NRS)

The idea of NRS is highly relevant to dynamics of mid-large-scale SESs. In these contexts, one cannot assume that there are distinctive or easily defined RSs and RUs. Rather, there can be multiple RSs, and each of these RSs can act as RUs depending upon the context within which they are being investigated. For example, let us consider a system represented in its entirety by multiple spatially connected lakes. Traditionally, we would imagine the entire lake system to be the RS and individual lakes within that system to be RUs. However, each individual lake is also capable of providing RUs such as fish, water, or pasturage on its own, thus allowing it to simultaneously function as a RS. Resource flows in this system can occur through multiple pathways—within an individual lake (for example, pasturage or harvesting fish from a lake), between two individual lakes in a network (for example water overflowing from one lake into the next via channels connecting the two), or across the entire network of lakes (for example, a system of water flows or the mobility of fish across the entire network). RUs too may be drawn at any of these levels—one can withdraw water from a single lake or across the system, while fishing or grazing livestock can occur only at the level of individual lakes. If we were to imagine the entire network of lakes to be our RS and individual lakes to be RUs, this distinction is not captured effectively. To address this discrepancy, we propose the idea of the nested resource systems (NRS)—conceptualised through Fig. 1. Highly relevant to mid-large-scale systems whose boundaries are not so easily defined, we propose that NRS may be considered as a complex of several individual semi-autonomous subsidiary RSs that may or may not be connected through physical connections. Each subsidiary RS can both provide RUs from the perspective of the NRS but is equally capable of acting as a standalone RS (thus semi-autonomous). There are system connections between different subsidiary RSs. These may come about by biophysical structure such as elevation gradients, or social structures like supply chains when seeds are bred, multiplied and sold as farming input in plant breeding systems. Activities, embedded in ASs can occur separately or simultaneously at four different spatial locations: (a) within the subsidiary RSs, (b) across individual RSs, (c) between the subsidiary RSs and the overall NRS, and (d) within the overall NRS. These ASs can occur across different levels of the NRS, and the outcome of any AS is likely to influence other ASs at any level of the NRS, causing spatially significant adjacencies. Actors operate across the NRS, leading to NAASs, where outcomes of individual ASs occurring at any one level can influence and be influenced by other ASs occurring at other levels. It is important to note that the NRS is situated within its biophysical environment and has multiple (not necessarily linked) components which when taken together define the SES.

Fig. 1
figure 1

Structure of an NRS with NAASs operating within it

Full size image

The diagnostic approach

To operationalise the concept of a NRS in the context of studying diverse SESs, we first developed a diagnostic protocol to logically unpack different components of a NRS and the NAASs within them that lead to SES outcomes. This diagnostic process was developed through a series of iterative discussions and deliberations amongst the research team drawing on our varied expertise and contextual knowledge of diagnostic protocols and mid-large-scale SESs. Like its use in healthcare, a diagnostic approach can tease out complexity within a SES as well as address the panacea problem (Frey and Cox 2015; Young et al. 2018). It allows the researcher to examine individual characteristics of a problem to identify governance arrangements that may best be suited towards addressing those characteristics (Young 2002, 2010). Diagnosis identifies underlying causes of a problem and works on the principle that addressing the problem would require intervention at causal levels (Cox 2011). It typically involves asking and answering a series of questions about the system such that subsequent questions build upon and add to information presented by previous ones (Berkes 2007; Ostrom 2007; Frey and Cox 2015).

Like the SESF, diagnosis allows typological decomposition of a complex system into its individual components allowing the researcher to unpack non-linear webs of relationships built by individual variables in SESs. It allows the construction of multi-level theories guided by similarities and differences between systems at multiple levels of specificity (Frey and Cox 2015). Such theories can then be used to provide some degree of generalizability and predictability to generate useful prescriptions on interacting with complex SESs (Cox 2011) and in the longer term, enable the creation of middle range theories.

Hinkel et al. (2015) establish a diagnostic procedure by providing a sequence of questions to facilitate stepwise interpretation and application of the SESF across diverse cases. The approach as outlined by them serves as a valuable starting point for this paper due to the following reasons. First, the approach explicitly focuses on RS and RU, due to their role in facilitating focal ASs and therefore the starting point towards applying the SESF to a given case (McGinnis and Ostrom 2014). Hinkel et al. (2015) advance the idea that the appropriation of an AS is inclusive of actors performing activities that depend upon a common stock and further subtract from it. Thus, the diagnostic tool they propose explicitly focuses on activities affecting the RU, allowing for the diagnosis of complex conditions within the SES such as multiple, overlapping, and heterogeneous actors and governance regimes. This is important because it has been shown that defining a stock as a collective good is not very effective largely because a stock by itself is not subtractable—it only becomes subtractable in relation to the activity associated with it (Hinkel et al. 2015). Subtractability as a characteristic is therefore only relative to the activity being performed in relation to the SES, while the property of excludability is related to actors performing the activity.

We acknowledge that ASs can take a wide range of incentive structures within natural resource governance and can include forms of cooperation, conflict, or indifference  (Bruns and Kimmich (2021) characterise these through a game theoretical approach as win–win, discord, and threat, with exchange, coordination, and independence as their primal archetypes), and it remains up to the researcher to determine the nature of the incentive structure associated with the SES challenge they are investigating.

Our diagnostic process builds on these premises and begins with identifying broad social ecological challenges that the analyst wishes to address, the research question as relating to the identified challenge/s and the specific SES or NRS that they engage with. We then provide a schematic that guides the analyst towards identifying assumptions behind the outcomes they envision, and a series of questions designed to unpack the complexity of RS and RU variables within the identified SES or NRS in relation to the research question they have identified. The schematic follows on to guide the analyst towards identifying NAASs, the spatial dimensions within which they occur, and outcomes that are generated as a result, all within the system of interest, bounded by the research question and level of analysis. These outcomes are then linked to the research question posed by the researcher, external influencing variables, and further on back to the broad SES challenge/s that they have engaged with. At various stages of the diagnostic process, we provide checkpoints that allow the analyst to ascertain whether their case study may be interpreted using the frames we provide. We do this so that focus remains on outcomes relating to SEASs that occur within SESs/NRS. Figure 2a–c outlines the diagnostic procedure we follow towards analysing and interpreting our cases. We exclude purely social outcomes from this diagnostic process because our focus is on NRS and changes within the RS usually occur because of social–ecological or purely ecological processes. Of course, if one were to consider governance systems and actors who form other first-tier components of the SESF, social outcomes become equally important drivers of social–ecological outcomes. However, for purposes of clarity in this diagnostic, which unpacks nestedness of RSs, we are excluding other first-tier components of the SESF. Thus, when listing out ASs, even though we use the typology provided by Schlüter et al. (2019a; b), we focus on SEASs as occurring within our NRS and its subsidiary RS. Our diagnostic tool (see Fig. 2a–c) is built keeping in mind the fact that multiple activities can contribute to one AS. For a step-by-step direction on how the diagnostic process may be applied to individual cases, please see Appendix 1.

Fig. 2
figure 2figure 2

a Diagnostic procedure: Section 1. b Diagnostic procedure: Section 2. c Diagnostic procedure: Section 3

Full size image

Cases and methods

We next tested the efficacy of our diagnostic process on unpacking SES/NRS and their associated NAAS dynamics across two distinctive and well-studied empirical cases. Our two cases represent distinct kinds of NRS: on one hand are networks of lakes, representative of traditionally studied common pool resources (other examples include fisheries and irrigation systems). On the other hand, we engage with German winter wheat breeding systems representing non-traditional, technologically mediated SESs (other examples include bioenergy and climate systems). Breeding systems differ from farming systems, as the underlying RU is the flow of genetic differences contained within physical material, like seed or plant parts (Gerullis et al. 2021), thus simultaneously making them divisible packages on a lower level (individual varieties or genes) and continuous streams of material on higher levels (maintained resistance to pathogens over time).

Plant breeding systems therefore show both characteristics of what McGinnis and Ostrom (2014) define as social–ecological technical systems (SETS). First, people dependent upon these systems view services derived from it as continuous streams (unlike traditional SESs where services can be obtained in discrete packets—for example, yields of fish from a lake). In wheat breeding, the benefits are measured through continued selection and propagation of the most suitable varieties for a geographic region. The second distinguishing characteristic of SETS is that there is often clear separation between actors possessing technical expertise to understand construction and maintenance of the system, and those whose sole concern rests with continued access to the resource (McGinnis and Ostrom 2014). In German winter wheat systems, clear separation exists between laboratory and field research stations (providing technical expertise) and commercial wheat farmers (who only depend upon continued access to favourable varieties).

We draw upon our long-term empirical research (see for example Castán Broto et al. 2021; Unnikrishnan et al. 2020; Unnikrishnan and Nagendra 2020; Gerullis et al. 2021) conducted in these contexts. The empirical documentation of NRS landscapes presented in Appendix 2 draws on data obtained through mixed methods approaches. In Bengaluru, these involved transect walks around 24 extinct and extant lakes to document the diversity of tangible and intangible benefits. Discourse analysis of archival documents (from CE 1800 onwards and from various sources: The Karnataka State Archives, The Mythic Society, and the Indian Institute of World Culture in Bengaluru; the National Archives in New Delhi; and the British Library in London) was used to generate historical data on benefits alongside oral history interviews conducted with elderly, long-term residents occupying a radius of 500 m surrounding the field sites. These data were combined with visuals of topographical change tracked through digitising battle plans and toposheets (from 1857, 1935, 1973) and on Google Earth (present day). For the German seed system, we used qualitative interviews, participant observation, and secondary sources from scientific literature or practical guidebooks on breeding, farming, and seed multiplication. Data collection followed a grounded theory style process; interviewees’ claims were anonymized, fed into modified questionnaires, and presented to subsequent interviewees for comment. Through this iterative approach, we consolidated individual perspectives into a knowledge consensus of the plant breeding system. To account for survivor bias and sequentiality, these consolidated accounts were presented to the first round of interviewees for validation in a final feedback loop. 18 interviews and 21 participant observations were conducted throughout 2016–2017, and 2019. We used participant observation to supplement our interview data with practical, first-hand experience of processes in breeding programs.

Case applications of diagnostic approach

Before moving on to the case applications of our diagnostic process, it is important to highlight an important consideration here. The delineation of system boundaries as well as the broad SES challenge/s within it relate to the specific research question being addressed. This distinction recognises that a system can be conceptualised in multiple ways and studied through multiple framings; however, it is up to the researcher to choose which framing is most useful for the purposes of answering the research question they originally set out to explore. We now demonstrate the applicability of our diagnostic process, as exemplified by the two case studies.

Networks of blue urban commons in Bangalore, India

Understanding drivers of coproduction around urban commons (Q.1.1 of diagnosis as presented in Fig. 2a–c; and Appendix 2) (such as lakes, parks, gardens, etc.) such that they produce ecologically grounded and socially just outcomes has been a long-acknowledged SES challenge. The case of networked lakes in south Indian Bengaluru is a good example of a traditionally studied SES, which easily lends itself to the idea of a NRS. Landlocked, situated in a rain shadow, and devoid of a major water source such as a river, the city has been unusually prosperous since ancient times and has served as a strategic location for colonial establishments (Unnikrishnan et al. 2020). This success of the city is partly attributed to a series of engineered water bodies dating back to about the fourth century CE which provided water to the city. These rain-fed reservoirs (tanks, lakes, or keres) were built by tapping into the city’s elevation gradient and utilising its naturally undulating terrain. Individual reservoirs were connected by channels, creating an engineered system of flows. Originally constructed to support agrarian communities, these reservoirs over time became integral to the cityscape, providing critical urban ecosystem functions and benefits (Unnikrishnan et al. 2020). This system of engineered water bodies transformed into novel ecosystems (Unnikrishnan and Nagendra 2020), characterised by complex interactions between society and nature, in turn functioning as complex social ecological landscapes.

Urban lake networks provide several shared long-term benefits—these include microclimate regulation, supporting resource dependent livelihoods, acting as biodiversity hotspots, and recharging shallow groundwater reserves. At the same time, given increasing pressures of urbanisation, and the landlocked character of Bengaluru, these reservoir systems have increasingly been viewed as a fluid conduit for the city’s sewage—a way to flush out wastewater from the city and into neighbouring regions. Lakes and their channels have also been seen as easily appropriable spaces to convert into other public infrastructure (malls, bus stands, and stadiums). Surviving reservoirs have either lost connectivity in parts of the chain or are treated as standalone water bodies where systemic connections are overlooked. We therefore have multiple social dilemmas arising in this context (Q 1.2). An overarching one relates to the maintenance of connectivity between individual reservoirs of the SES versus conversion of these spaces into other forms of built land use. A similar social dilemma is presented at the level of individual lakes within the network: the maintenance of individual water bodies versus their conversion into built structures or their reimagination as primarily economically driven entities (Unnikrishnan et al. 2016). Individual lakes provide similar and relatively long-term ecosystem services as the larger network—microclimatic regulation, biodiversity, support for resource-based livelihoods, and serving as a local water reservoir. At the same time, in the short term, they are attractive prospects either for redevelopment as real estate or to enhance the value of existing real estate by providing aesthetic and recreational services (Unnikrishnan et al. 2016). This latter viewpoint brings with it several social–ecological challenges: converting lakes into built up spaces increases the risk of urban flash flooding, creates social vulnerabilities among resource dependent populations, and reduces diversity of ecosystem services they provide. However, the larger trend in the region seems to be driven towards an aesthetic and recreation dominated urban vision (Unnikrishnan et al. 2016)—a vision that seems to have sustained itself across at least two centuries.

Considering this contextual background, the objective of applying our diagnostic process is to understand what motivates co-production in this network of lakes in Bengaluru? In other words, what drives inherently heterogeneous communities to invest in the resource collectively? Normatively, we seek to understand what factors may influence heterogeneous communities to engage in lake restoration such that one may achieve favourable ecological outcomes (such as improved water quality or biodiversity) alongside socially just ones (such as representation of diverse interests in the resource) (Qs: 1.3–1.5).

As the network of lakes consists of several individual lakes connected through channels, each of which in turn provide various social–ecological benefits, this system is representative of an NRS. The broad NRS is represented by the network of lakes, while individual water bodies within the network form semi-autonomous subsidiary RSs. Each semi-autonomous subsidiary RS can act as an RU within the NRS but is equally capable of providing RUs (such as fish, water, etc.) by themselves (Qs 1.6–1.9). Actors within this NRS are represented by internally and externally heterogeneous groups of people drawing tangible and intangible benefits—ecosystem services (MEA 2005)—from the resource. Provisioning ecosystem services (material and quantifiable benefits obtained from the system) take the form of entities such as water for commercial and subsistence uses, fish, urban forage, etc. The diversity of intangible benefits such as support for spiritual beliefs, community building, recreation, and aesthetics, are cultural ecosystem services obtained here. Of benefit to the general population and subsequent generations living around the lakes are various regulating services such as pollination, and microclimatic regulation, along with supporting ecosystem services such as nutrient recycling and biodiversity maintenance.

Farmers, fisherfolk, recreationalists, urban foragers, nodal agencies and various other actor groups undertake different activities in and around the NRS. Several actor groups draw benefits from the NRS, through varied activities that are regulated in multiple ways (see Appendix 2 for detailed listing of actor groups and institutional arrangements). The number of actors undertaking these activities as well as ways in which these activities are regulated have implications for the subtractability of stocks of RUs (stock of fish or number of entire lakes), in relation to the activity (fishing or draining entire lakes for building), as well as how easily excludable other actors are from conducting the same activity. These may influence the availability of various ecosystem services (Qs 2.1–2.7).

Figure 3 exemplarily illustrates various activities that give rise to ASs, which occur at multiple levels of the network. Some ASs occur only at the level of the individual lake, whereas others, while taking place at individual lakes, can be influenced by activities occurring elsewhere across the network. For example, the AS characterised by occupying spaces around lake banks and associated fishing activity takes place at the level of individual lakes within the NRS. At the same time, fishing is dependent upon proper functioning of systemic connections across the network. The availability of fish within individual lakes, as well as the quality of water supporting those fish, are both characteristics of the SES that are dependent upon RU flows across the network. Thus, this AS, while occurring at the level of an individual lake, remains deeply embedded within system dynamics of the larger network that it belongs to. This is different from the AS involving appropriation of pasturage from banks of lakes to support livestock grazers, which necessarily occurs only at the level of individual lakes—its functioning remains relatively independent of activities occurring within the broader network (Q: 2.8).

Fig. 3
figure 3

Exemplary illustration of the network of urban lakes as an NRS with NAASs operating within

Full size image

These ASs do not exist independently of each other however, and there are several adjacencies which are created. For example, ASs involving privileged gated communities who appropriate land around individual lakes for real estate, almost always are linked to ASs involving local nodal agencies who are responsible for maintenance and governance of the entire network of lakes. Similarly, ASs involving the appropriation of land and water in and around an individual lake (which are themselves influenced by the larger network that they are part of) are linked to those involving appropriation of pasturage from around individual lakes, largely due to the association between agricultural practices and livestock rearing in the region. What this means is that adjacencies are not created simply between two ASs, but that they can occur along different spatial levels within the NRS. Each of these interactions further link themselves to social–ecological outcomes—in this case with its explicit focus on motivations for co-production, we define these outcomes by the ecosystem services that are enabled or disabled within the system (Qs: 3.1–3.2).

In applying our diagnostic process to this case (Appendix 2), we find that only four user groups (nodal agencies, gated communities, private institutions, and urban recreationalists) possess all the following attributes: (a) access, appropriation, management and/or exclusion rights; (b) despite being affected by the larger lake network, tend to operate at the level of individual lakes; (c) access to stakeholder collaboration and information flow; and (d) the ability to directly influence form and function of the ecosystem, while accessing only cultural ecosystem services. This means that power to influence the SES is monopolised by these groups of actors, providing them with greater incentive to engage in co-production efforts towards the resource. Ecologically, this means that efforts are not systemic but targeted only to individual lakes, meaning that the entire NRS is not sustainably rejuvenated.

There are other actor groups who only have access and appropriation rights, are more diverse, depend mostly on provisioning ecosystem services, and who in some cases draw meaning from the systemic nature of the resource. However, these groups usually do not influence the condition of the resource and are not involved in decision-making processes around it. Hence, there is very low incentive for these actors to come together and engage in co-production efforts. This implies that in this case, the success of coproduction around blue commons seems intimately linked to how inclusive the process is to diverse stakeholders of the NRS (Qs: 3.3–3.5).

German winter wheat breeding systems

We utilise our diagnostic process to answer what governance challenges arise in appropriating and provisioning ASs for crop genetic diversity in German winter wheat breeding systems. Relevant for answering this research question therefore is a combination of social and biophysical outcomes. We need to know whether (a) breeders are creating varieties maintaining their long-term genetic pool; (b) subcontracting and selling varieties such that farmer’s needs and preferences are being met; (c) farmers are choosing varieties according to their own ecological niches and preferences, such that negative ecological and societal impacts are minimised (Qs:1.1–1.3).

Plant breeding systems (“breeding systems” subsequently) are good examples of SETS involved in creating, maintaining and improving seeds of different crop varieties for farmers to produce food and fibre for human use. Aside from the usual resources used in farming like land, water, fertiliser, and chemicals, breeding systems depend upon genetic variation contained in different plant materials used for breeding. These are very diverse and range across physical material from single allelic snippets, seed, other plant parts, and single plants to variant plots and fields. For actors involved in breeding activities (breeders, plant scientists) these physical flows coincide with information for observing genetic differences in these materials—called traits. As one can tell from this inherent nestedness, these are also NRS. Plant breeding systems as nested, multilevel systems supply and provide affordance for different flows of genetic material in any form and its corresponding information.

We refer to the overall stock of these traits as “genetic diversity” in the following. For actors in seed multiplication, retailing and farming, relevant information results from differences in bundles of traits, called varieties. We refer to the overall stock of these trait bundles as “varietal diversity” in the following (Qs: 1.4–1.5).

The diversity of actors here includes scientists working in crop science or pre-breeding, breeders/breeding firms, seed multipliers, different governmental and non-governmental organisations, and farmers, who plant the varieties and provide their harvest to the system for processing food and fibre. Breeding systems are nested in their underlying genetic set-up along pedo-climatic zones, for which pre-breeding and breeding actors develop varieties (Q: 1.6).

There are economic benefits, mainly income, created for all actors along the seed supply chain: income is generated from variety licensing and subcontracting, selling seed and sales of other inputs accompanying seed (crop protecting agents, fertiliser, machinery). Farmers sell their yield and as such security from stable yields over the years is also a direct economic benefit. Other benefits generated by the system are the future value of a genetically diverse system, and may entail ecosystem services touched by agriculture, like nutrient cycling, groundwater quality, pollination and biodiversity maintenance. The benefits are created from multiple levels within the NRS. While scientists are changing the RUs on a molecular level, applied breeders are interested in changing whole plants, farmers sow seed on the level of their farm plots, retailers push for sales across regions, and governance organisations care about the multiplication areas in regions and states (Qs: 1.7–1.9).

A social dilemma in the appropriation of genetic diversity occurs when breeders reduce the genetic variation in their used material to the point where their gene pool does not contain certain needed traits to maintain cultivar yields anymore, e.g. resistance against a fungus. This may occur when breeders focus their breeding practice on mainly “low hanging fruit” such as qualitative resistance traits. As qualitative traits are determined by only a few allele sequences in the DNA, there is less delay in progress when establishing a new trait into a new variety candidate. Modern molecular marker technologies will allow breeders to find these at a low cost once they are identified. Thereby they can be easily combined into new varieties. Focusing on qualitative traits, nonetheless, comes at the expense of more complicated traits, as there is a trade-off between different breeding goals. If breeders decide to put more resources towards breeding resistance traits they cannot pursue other goals with equal power, as breeders are restricted in their nursery space, person-power, and nursery locations within different environments. A reduction in complex resistance traits would reduce genetic diversity negatively across all breeders. Maintaining genetic diversity of all kinds of traits, however, is vital for sustaining breeding activities and agricultural systems in the long term.

A social dilemma in the appropriation of varietal diversity emerges on a higher level. When too many farmers plant only one variety over large areas, new strains of pests can evolve thereof. One example of this is the occurrence of European yellow rust epidemics in winter wheat of recent decades (Bayles et al. 2000), where strains of plant pathogens evolved from overuse of susceptible varieties. To counteract pests, farmers spray pesticides to prevent the risk of yield losses. Yet, farmers end up spraying more pesticides than necessary (Dachbrodt-Saaydeh et al. 2018), leading to unwanted externalities in their natural environment (Qs: 3.5–3.6).

There is a social dilemma that emerges overarching the two social dilemmas described above. If farmers revert to over-spraying for risk-reduction every year, they need not rely on choosing varieties with well-working resistance but will choose susceptible high yielding varieties (Dachbrodt-Saaydeh et al. 2018). This decreases the market share in varieties with well-working quantitative resistance traits and leads to an added disincentive for breeders to invest in costly generation of these traits (Qs: 3.5–3.6).

The objective of applying the diagnostic process is to understand how to maintain varietal and genetic diversity considering these perverse effects. For German winter wheat cropping, part of these effects is intercepted by governmental regulation and public information diffusion. This is enabled through extension services and public–private breeding efforts. For example, through pre-breeding programs, or encouraging social norms amongst breeders, in ways that reward breeders with the prestige of creating varieties containing complex traits. We are interested in how governing material and information flows on the different levels of the NRS bring about different outcome patterns in genetic and varietal diversity.

Breeders, seed multipliers, retailers and farmers undertake various activities (see Appendix 2, Sect. 2; Qs: 2.2–3.1), which change the shape and size of underlying resource stock of genetic diversity, where each activity is bound by different institutional arrangements, yielding a multitude of NAASs. For example, breeders’ activities will influence the stock of RUs of in-nursery genetic diversity and devise the available varieties for other actors. Institutional responses in collective norms on material exchange, state regulations on variety approval and open access to approved varieties influence how the social dilemmas are met. Excludability of actors from various activities is easy, difficult or in some cases varies by individual, as some enabling preconditions determine whether one can undertake the activity. Likewise, subtractability of the resource stock through activities may vary by individual actor or depend on heterogeneous spatial circumstances—for example, subcontracting of varieties for different regions depends on the ecological niches covered.

The earned and expected income gain incentivizes different actors to undertake the activities. Information flows on different agronomic performances of individual biological material (genetic snippets, lines, varieties) direct concrete genetic material to their purpose and spatial positions within the system, leading to different ecological performance measures. Figure 4 exemplarily shows that ASs are networked in two ways: first, through the nestedness of the RSs, where changes on one level of the RS entail changes in the RS on a higher level influencing patterns in ASs on that level. For example, a change on the molecular level of genetic traits about a resistance trait in a variety may impede pest outbreaks in fields of farmers. Second, through transactions taking place between different actors in the respective ASs, where breeders exchange breeding material containing resistance traits and produce varieties which are resistant to certain pests. Multipliers subcontract these varieties if they perform well and sell them through retailers to farmers. Farmers will only spray less if their varieties are resistant to all diseases relevant to their farm. These relationships are dynamic. The ecological world constantly evolves, where pests evolve resistance to formerly tolerant varieties, and plant research is racing to keep up with natural selection. Likewise, social mechanisms of market transactions, subcontracting and collective action change as wider economic and political settings change over time and exert comparable social selection pressures (Qs: 2.2–3.2).

Fig. 4
figure 4

Exemplary illustration of the German winter wheat breeding system as a NRS with NAASs operating within

Full size image

Three individual social dilemmas in networked ASs need to be overcome to not encounter negative environmental impacts on the overall system level: breeders need to invest collectively in quantitative resistance traits to have these in their varieties. Multipliers need to be willing to subcontract these resistant varieties and forego income from accompanying plant protection agents, so that farmers may sow varieties with stable resistances against pests and spray less crop protection agents. In all three of these ASs, the incentives each actor group is faced with point in different directions (Qs: 3.4–3.6).

Discussion and conclusions

In this paper, we engaged with two broad challenges of the SESF. First, we build upon the gap first articulated by Cox (2014) on insufficient decomposition of RS and RU. We attempt to formalise this within the structure of an SES by introducing the NRS—the idea that an RS can function simultaneously as both RS and RU depending upon framing of the problem at hand and the boundaries of the system that emerge because of that particular problem frame. We show how NRS contribute to NAASs as nestedness of the NRS leads to a biophysical connection between different ASs (Schlüter et al. 2019). Depending upon one’s inquiry, our diagnosis makes physical connections between different ASs visible, providing an opportunity to show these connections spatially, and thus making NAASs spatially explicit (Möck et al. 2019). Second, we propose a diagnostic tool to aid analysts in applying the SESF and articulating associated NAASs to their cases in a standardised manner, allowing for greater comparability across diverse cases (Kimmich et al. 2022). We thus take a step forward in the direction of addressing the acknowledged gap of establishing a protocol for NAAS research (Kimmich et al. 2022; Müller et al. 2013) We have tested the diagnostic process within the context of two distinct systems—a SES characterised by urban lake networks in Bengaluru and a SETS represented by German winter wheat breeding systems. We believe that this diagnostic process may be used successfully in unravelling complexities of other kinds of SESs such as knowledge commons or what are called “new commons”. In this section of the paper, we reflect upon the utility of these approaches in expanding our understanding of the SESF and its application to understanding environmental governance challenges.

Decoupling RSs and RUs brings distinct methodological advantages when applying the SESF to cases. First, it allows us to engage with complex dynamics of mid–large-scale systems where there is significant diversity of simultaneously occurring activities operating at multiple spatial levels. Second, it allows us to engage with fluidity of boundaries existing between RS and RU components, while understanding that identities of the RS and RU are largely dependent upon specific activities as opposed to being defined as fixed systemic characteristics (Hinkel et al. 2015). Third, given that RS and RU form the starting point for defining focal ASs, this decoupling allows us to incorporate consideration of simultaneity of interconnected ASs occurring across multiple spatial levels and leading to cumulative outcomes on the SES, which makes it representative of NAASs. It, therefore, provides a first step towards unpacking substitution effects and redundancies that emerge from the complex interplay between actors, their activities, and regulation of those activities.

The application of our diagnostic tool, following the deconstruction of RS and RU allows the analyst to unpack the SES in a standardised manner. This allows for comparisons across diverse cases through meta-analysis—the systematic and consistent coding of cases using the SESF, following which the analyst can observe for patterns of similarities and differences across cases (Ostrom and Cox 2010). These comparisons also provide useful data points for large N-case studies of NAASs and, therefore, serve as a base to aid the generation of middle range theories.

The two cases we analyse using this diagnostic tool help us outline some of these similarities and differences. Both cases are diverse in that they are representative of two distinct systems—an SES and SETS, that are difficult to compare across the geographies in which they occur. At the same time, in decomposing RS and RU components of these systems and applying the diagnostic process, several commonalities come to fore. First, there exist physical connections between RUs in the different ASs of each system—for example, the channels which connect individual lakes within the NRS (enabling flows of water) are comparable to the flows of genetic material enabled in the form of seeds. Second, both systems have multiple social dilemmas occurring at different scales—some of these form overarching dilemmas, while others restrict themselves to the subsidiary RSs in these systems. A complication that emerges from the presence of these multiple dilemmas is that overarching social dilemmas cannot be addressed without engaging with those that occur at lower scales of influence. This is further complicated by inherent heterogeneities emerging between actors, activities, and ASs at multiple levels of the NRS. Third, diagnosis brings out commonalities in the kinds of substitution effects emerging with respect to activities occurring within the NRS through a consideration of simultaneously occurring ASs. For example, the substitution of provisioning food activities in the lake NRS with increased aesthetic and recreational ones would mean that certain user groups are almost immediately excluded from decision-making involving the NRS. Fourth, both systems show redundancies in that multiple actor groups can perform the same activity, or that multiple identities are assumed by the same actor group, therefore with potential to influence the rules-in-use governing these systems. The presence of redundancies means that you can either be an all-in-one entity internalising harmful effects, or you have multiple redundant groups (for example, the farmer/livestock owner) with different abilities to negotiate harmful effects. In the latter case, negative effects are likely to be experienced by those excluded from decision-making either through negotiation or imposition by other groups, as has been demonstrated in both cases we analyse.

From a managerial perspective, these commonalities provide insight into critical points of intervention needed in NAASs occurring within NRS. From the perspective of lake networks, the analysis highlights the need to include information flows across all actor groups, especially those who engage with both systemic and individual levels of the NRS. For example, nodal agencies and real estate groups engage in decision-making around converting lakes, which influences the entire system, yet they do not include fishermen and farmers, who are affected by these decisions. A potential goal, therefore, is to reach stewardship for the lakes’ condition across all actor groups, as each individual group can through overuse, hamper social–ecological outcomes of the NRS. From the perspective of plant breeding systems, an important governance challenge involves public agents providing information of variety trials within and across actor groups depending on the level of the RS at which actors operate. For example, maintaining genetic diversity amongst breeders depends on common use and exchange of breeding material, in lengthy processes over several years, which obfuscate causal links between management decisions and breeding outcomes. Distributing information amongst breeders from trials preempts this process and minimises opportunities of defecting amongst breeders, while incentivizing individual improvements of the genetic pool. Yet, economic considerations of ASs later in the seed supply chain (NAASs), influence which material breeders use. Hence, a potential goal for plant breeding systems governance is that variety trials give information such that decision-making by breeders and farmers enables a choice of more sustainable strategies for choosing varieties and breeding material.

These insights on systems could have been generated by other means and using other frameworks. Using the proposed diagnosis, however, will nudge the researcher to explicitly illustrate (a) how the nestedness of an RS does influence decision-making of the actors (incentive structure), (b) how biophysical configurations and information flows arising thereof diverge or overlap for different actor groups. Explicitly showing connections or disconnects between configurations of RS and incentive structures can aid in development of context specific, feasible solutions.

There are some caveats to using this diagnostic process. Applying it to a case requires that the analyst already possesses embedded knowledge of the system. Our diagnostic process also does not intend to prescribe normative views of interactions and outcomes; rather it encourages the analyst to make their own normative assumptions explicit in the process of applying the SESF to a particular case through critical reflection. Our conceptualization of the NRS and its subsequent application in the diagnostic process restricts itself to RS and RU components of the SESF. Engagement with how NRS and NAASs interact across diverse GSs would be a very useful next step along with engaging explicitly with second-tier variables of the SESF. Similarly, advancing diagnostic tools to enable dynamic comparisons across temporally situated NRS and NAASs could enable better comparisons, aiding development of middle range theories drawing on institutional emergence. However, one should note that while this paper adds structure to complexity for individual cases, there is a trade-off between fleshing out detailed items within our diagnostic approach and linking this to data entry and meta-analysis. Future research should consider this trade-off alongside applying this diagnostic tool to multiple cases for comparison and meta-analysis, drawing upon the SESF and NAASs.