A systemic framework for context-based decision making in natural resource management : reflections on an integrative assessment of water and livelihood security outcomes following policy reform in South Africa

We aimed to contribute to the field of natural resource management (NRM) by introducing an alternative systemic context-based framework for planning, research, and decision making, which we expressed practically in the development of a decisionmaking “tool” or method. This holistic framework was developed in the process of studying a specific catchment area, i.e., the Sand River Catchment, but we have proposed that it can be generalized to studying the complexities of other catchment areas. Using the lens of systemic resilience to think about dynamic and complex environments differently, we have reflected on the development of a systemic framework for understanding water and livelihood security under transformation in postapartheid South Africa. The unique aspect of this framework is that allows researchers and policy makers to reframe catchments as being recognizable as complex socialecological systems, and by doing so, the possibility is opened to understand resiliency in the face of rapid transformation and crisis. Ultimately, this holistic approach can be used to understand the translation of policy into practice. We have emphasized our reflections on the development and use of the framework and the challenges and successes faced by collaborators in the process of adopting such an orientation. Because these are likely to characterize policy and decision-making processes in NRM in general, we have suggested that such a systemic framing can assist researchers, practitioners, and policy makers to adopt systems and resilience analyses in the process of planning and implementation.


INTRODUCTION
There is evidence of a growing discomfort with governance and management based on linear cause-and-effect paradigms, often supported only by reductionist science (Forrester 1992, Holland 1999).Alternatives to these orientations are emerging in many sectors from natural science (Gunderson et al. 1995, Meadows 1999, Folke et al. 2002, Walker and Salt 2006) to business (Snowden 2000), education (Forrester 1992, Ison et al. 2007, Wals 2007), service delivery by government (Radzicki and Taylor 1997), and in disaster relief (Ramalingam et al. 2009).These concerns are supported by philosophical and epistemological critiques by scholars such as Edgar Morin, Paul Cilliers, and others (see Nowotny 2005, Heylighen et al. 2007).
Although alternative approaches are emerging from such diverse interests, the development of alternatives that embrace nonlinear cause-and-effect paradigms have been slow in the field of natural resource management (NRM).We offer an alternative approach that contributes to a systemic framing for planning, research, and decision making in complex environments.The framework is based on experiences from catchment-based work that started in the early 2000s regarding water governance reforms and the adoption of integrated water resources management (IWRM) in postapartheid South Africa.Although the work is ongoing, we reflect on a three-year study from 2005 to 2008 that sought to understand the impacts of policy-related transformation specifically on water and livelihood security in the Sand River Catchment (SRC), but to do this in a way that acknowledged the complexities and uncertainties typical of the management of catchment areas in general (Pollard et al. 2008).By viewing the catchment as a complex, dynamic social-ecological system (SES), we sought to examine the coupled nature of degradation, vulnerability, and resilience, i.e., social and biophysical, so as to explore the potential multiple outcomes, and lags, of policy reform as it plays out in complex environments.
The unique aspect of this framework is that it allows researchers and policy makers to collectively reframe catchments as recognizable complex SESs, and by doing so, the possibility is opened to understand resiliency in the face of rapid transformation and crisis.Although the concept of resilience was incorporated (sensu Berkes et al. 2003) when the study started, work on applying resilience in practice was still under development, notably, the resilience workbooks (RA 2007a, b), so that the work we report took place in parallel with those developments.Although the approaches share many features, some distinctions are elaborated.
Our emphasis is on our reflections on the development and use of the framework.Thus, the focus is not on the primary results reported in Pollard et al. (2008), although some illustrative exemplars are given, but rather to probe the extent to which such approaches can offer an effective way of dealing with systemic complexity in similar contexts.It is suggested that such a systemic framing can assist researchers and policy makers to adopt systems and resilience analyses in the process of planning and implementing policies dealing with water and livelihood security.
We define one of the two central concepts, water security, as "sustainable access, on a watershed basis, to adequate quantities of water of acceptable quality, to ensure human and ecosystem health" (Norman et al. 2010:10).The second concept, livelihood security, derives principally from the work of Chambers and Conway (1992), and as a component of this, household livelihood security is taken as "adequate and sustainable access to income Ecology and Society 19(2): 63 http://www.ecologyandsociety.org/vol19/iss2/art63/and resources to meet basic needs (including adequate access to food, potable water, health facilities, educational opportunities, housing, and time for community participation and social integration" (Frakenberger and McCaston 1998:31).Within this, access to water, for human and livestock needs, as well as for food production and other economic activities, which is the focus of this work, is an obvious contribution.
We begin with a brief description of the national policy and local catchment context, as well as the conceptual framing underlying the approach.Practically, the framework manifests as a suite of 10 inter-related steps presented in the Methods.The results focus primarily on our reflection on the use of this approach, although some case detail is provided to illustrate how the approach was used and the key outputs, which are discussed in relation to the wider implications for planning and management processes.We conclude by reflecting on some of the challenges and successes that the collaborators encountered in the process of adopting such an orientation and that are likely to characterize policy and decision-making processes for water governance in dynamic situations in general.

South Africa: a country of transforming policies
Political transformation, such as that experienced in South Africa with the transition from apartheid to democracy, can provide windows of opportunity to change such linear orientations and perspectives.In 1994, most South African policies underwent a major overhaul under the first democratic government.Social grants were introduced to support the vulnerable, and education was reformed.In the natural resource arena, water received concerted attention because of the country's chronic water insecurity and disparities in access (DWAF 2004b).Equity and sustainability were key principles of the new water policy, and, in recognition of water's pivotal role in socioeconomic development, holistic approaches such as IWRM together with cooperative governance and stakeholder involvement are now central (Schreiner and Hassan 2010).Although overall IWRM falls under the national minister, water governance is decentralized, with domestic water supply a municipal responsibility and delegation of water resources management to nine catchment management agencies (CMAs), which are in the process of being established.
The new national water acts, i.e., Water Services Act No. 108 (Republic of South Africa 1997) and National Water Act No. 36 (Republic of South Africa 1998), pertaining to water supply and management were promulgated in 1997 and 1998, respectively, and implementation was begun in earnest.In the many waterstressed areas, questions soon emerged regarding the extent of biophysical degradation and socioeconomic vulnerabilities, such as the role of water in poverty alleviation, and what remedial actions were necessary.A decade after promulgation, questions were arising concerning the seemingly slow pace of change and impacts of policy on water and livelihood security and whether resilient futures were being built.Despite ostensibly enabling policies and emerging institutional arrangements, there often appeared to be little wide-scale change on the ground.Single cause-and-effect analyses and responses used, e.g., building more dams, were starting to be regarded as overly simplistic in dealing with real-world realities and complexities, and more holistic and innovative ways to understand the continued water and livelihood insecurity were thus being sought by some.In many cases, the application of questions regarding durability and unintended consequences pointed to systemic failures of individual interventions.One such area of concern is in the SRC in the far northeast of the country, which thus offered an opportunity to be reframed as an SES with resiliency providing the capacity for renewal, innovation, and stability (sensu Holling 2000) in the face of rapid transformation and crisis (Berkes et al. 2003) The Sand River Catchment: an area under transformation?Our focus concerns the SRC, which is located in the northeastern part of South Africa (Fig. 1).Nonetheless, it encapsulates many of the issues and challenges faced by other catchments in South Africa as moves are made toward transformation for more sustainable and equitable futures.It is widely regarded as degraded and vulnerable, particularly because already stressed water resources are under pressure to meet further developmental demands (Pollard andDu Toit 2011a, Pollard et al. 2011).The area exemplifies many densely populated rural settings where juxtaposed characteristics of wealth and poverty, ecosystem health and degradation, and increasing contestation around natural resources together test giving effect to the principles of equity and sustainability.Like many catchments, the imperatives to generate and share wealth through redress of racial discrimination, land reform, and development must be balanced with long-term environmental security.Institutional arrangements are currently in transition, highly dynamic, and often confusing on the ground (Pollard et al. 2011).
Relatively small in area (2000 km²) and home to some 383,000 people, the SRC is regarded as severely degraded biophysically, and vulnerable and underdeveloped from a socioeconomic perspective (Pollard et al. 1998).It forms part of the Incomati Basin, an international watercourse, the South African portion of which forms one of 9 legally constituted water management areas.With the exception of the wetter, western mountainous region, the catchment is semiarid and is increasingly in water deficit, i.e., water demands exceed water availability, with the result that the once perennial Sand River now experiences flow cessations in some dry years (DWAF 2004a, Pollard et al. 2011).
For the purposes of analysis, we used the three broad zone areas (A, B, C) described by Pollard et al. (1998) that had emerged by the mid-1960s (Fig. 1).The three zones represented an afforested higher rainfall upland zone; a densely populated zone with intermediate elevation and rainfall, stock, and crop farming often at subsistence level; and a more arid western zone dominated by private and state conservation areas and ecotourism.These zones also reflect politico-historical factors that influenced land tenure and land use.Although each zone represented relatively homogeneous land uses, systemic links are evident in the pejorative downstream impacts of land use in Zone A and in the labor and cash flows from and to Zone B.
A striking feature today is the dense concentration of people (350 per km²) in so-called rural areas of Zone B juxtaposed with the sparsely settled, often affluent areas of Zone C, which are dominated by conservation.These differences reflect the legacy of apartheid where large numbers of people classified as "black" were forcibly moved into two former bantustans, i.e., land set aside for the exclusive occupation of black people, which made up Zone B, where levels of unemployment, illiteracy, and poverty were http://www.ecologyandsociety.org/vol19/iss2/art63/Fig. 1.Map of the Sand River Catchment, showing the three major zones that comprise the catchment.high, and livelihoods vulnerable (Beinart 2001, Pollard et al. 2008).Environmental degradation increased as people turned to natural resources to make or supplement a living (Shackleton et al. 1995, Pollard andDu Toit 2013).Males migrated to the urban areas in search of work, and female-headed households were prevalent (M.Collinson, S. Tollman, K. Kahn, and S. Clark, unpublished manuscript).Some government agricultural schemes were established in Zone B in an attempt to address the high levels of unemployment and poverty.Although beset by lack of access to water and other resources, these continue today and support between 600 and 1000 small-scale farmers, each with 1 ha of land.The high-altitude Zone A, where more than half of the catchment's water production occurs, was dominated almost exclusively by forestry, i.e., afforestation, mixed with indigenous forest and grassland (Pollard et al. 1998).Forestry enterprises were established in the 1970s in an attempt to provide employment for the burgeoning population in Zone B, but multiple perverse incentives rendered this activity largely uneconomical.In an attempt to meet contractual commitments for timber, vulnerable areas, such as steep slopes, riparian zones, and wetlands, were cleared for plantations.The impacts were felt through reductions in base flows to the Sand River and increased sediment production and transport (Pollard et al. 1998).Ambitious and visionary attempts to transform this situation post-1994 through zoning the area as a national park and the removal of forestry have stalled more recently as political commitments and interests have changed.Zone C, which includes both the Kruger National Park and the exclusive Sabi-Sand Game Reserve, has enjoyed relative economic prosperity as both tourism and land values have increased in the past 15 years.Interestingly, although economically powerful, as downstream residents of the catchment, these players have experienced growing water insecurity leading them to participate more broadly beyond their own fences (Pollard et al. 2003).
Changes in policies regarding NRM have been more nuanced through the strengthening of protected area policies, stewardship initiatives, and massive job creation programs, epitomized by Working for Water, meant to remove alien invasive plants with the aim of improving water supply, which became operationalized and commonplace, and all very visible in the SRC.Many other reforms influenced the catchment, the details of which are beyond the scope of our discussion but which included major educational and health care reforms, land tenure reform, restitution for the dispossessed, and the introduction of child-care grants for the indigent.Noteworthy among new drivers were the increase in HIV/AIDS and the pejorative impacts thereof (see, e.g., Hunter 2007).Nonetheless, the seemingly slow pace of change raised concerns regarding transformation.Despite enabling policies, there appeared to be a number of "sticking points" or blockages and lags in the SRC, which were being dealt with through approaches that we hypothesized failed to recognize the systemic nature of water security.

Conceptual framing
The key conceptual underpinning of this work is that of systems thinking.Systems thinking (Von Bertalanffy 1972, Checkland 1981, Forrester 1992) includes theories that concern themselves with complex phenomena; such theories arose partly as a critique to conventional reductionist approaches, which were considered to be ill-equipped to deal with complex interdependencies such as those found in NRM.The overall thrust in dealing with complex phenomena is to foster a broader view of overall context, challenging notions of optimization, maximum sustainable yield, and linear thinking (see, e.g., Gunderson et al. 1995, Cilliers 1998, http://www.ecologyandsociety.org/vol19/iss2/art63/Levin 1999, Holling 2001, Folke 2003, Walker and Salt 2006).As stressed by Meppem and Bourke (1999), conventional NRM unrealistically abstracts, usually unidisciplinary, interests from real-world complexity.
Focusing on the complex inter-relationships among constituent parts, and thus on the whole through systems thinking, is a complement rather than an alternative to specialized views.Indeed, systems approaches incorporate both systemic and systematic perspectives (see, e.g., Laszlo and Krippner 1998).Important concepts contained in systems approaches include interdependence; holism and emergence; goal-seeking behavior; feedbacks and regulation; hierarchy; differentiation; equifinality, alternative ways of attaining the same objectives, i.e., convergence; and multifinality, attaining alternative objectives from the same inputs, i.e., divergence.These concepts are reviewed by various authors (see, e.g., Cilliers 2000).
Work on an integrative theory for coupled human-ecological systems, an SES, had culminated in a book entitled Panarchy: Understanding Transformations in Human and Natural Systems (Gunderson and Holling 2002).Panarchy was a term coined to describe the structure in which systems, e.g., SESs, are interlinked in never-ending adaptive cycles of growth, accumulation, restructuring, and renewal, known as the generalized adaptive cycles (GACs).These transformational cycles take place in nested sets at different scales.The authors suggested that by understanding these cycles at multiple scales, it seemed possible to identify points at which a system would be capable of inducing change that could be used as leverage points to foster resilience and sustainability deemed positive to stakeholders.Earlier work by Holling (2000) had also suggested that in trying to understand complex, evolving systems, there is a requisite level of simplicity that, if identified, can support understanding that is rigorously developed but that also can be lucidly communicated.He argued that if one cannot retain a handful of causes in an explanation, then the understanding is simplistic; whereas if more than a handful of causes are elaborated, then it is unnecessarily complex.That level of understanding is built on a sound integrative theory, rooted in empirical reality, and communicated clearly with metaphor and example.This integrative theory has been further developed through the closely related concept of resilience, which broadly refers to the capacity of a system to absorb disturbance and reorganize so as to retain essentially the same function, structure, and feedbacks (see Berkes et al. 2003).Indeed, the Resilience Alliance (RA; http://www.resalliance.org)has popularized the handling of complexity through the exploration of resilience based on the central tenet that because variation absorbs shocks and confers resilience it should be embraced, not ignored.Further, a focus on resilience shifts the attention from purely growth and efficiency to recovery and flexibility and supports learning and adaptation.Ongoing work has asserted that a number of attributes confer resilience including feedbacks, diversity, innovation, polycentric and overlapping governance, social capital, ecological variability, openness, and reserves (Walker andSalt 2006, RA 2007b).
Our discussion encompasses these ideas and focuses particularly on feedback loops and their role in systemic issues associated with NRM.For example, feedbacks, often operating at different scales, cause emergence, i.e., the feedbacks generate surprising new properties not predictable from the original components making up the system.In feedbacks, an output from an event or phenomenon in the past influences an occurrence of the same in the present or future (Holland 1999).Understanding feedbacks also proved central to exploring the so-called lock-in traps, i.e., situations in which the adaptive cycle becomes "stuck" at one particular point and cannot continue its normal cycle of change, described by Allison and Hobbs (2006) that have led to continued degradation despite changes in policy and practice.More recently, Pollard and Du Toit (2011b) have suggested that multiscale governance feedbacks are essential for supporting resilient IWRM systems.
When considered together, these theoretical framings suggested that by adopting a systems view based on the notion of catchments as coevolving, complex systems, it might be possible to understand cyclical transformation and the leverage points of Gunderson and Holling (2002), as well as the factors that confer or undermine resilience.In effect, such endeavors form part of a broad body of work on resilience assessments, which elucidate how linked SESs respond and adapt in the face of disturbance such as changes in land use (Walker et al. 2002), identifying key social and ecological variables and thresholds that determine system status.This helps develop strategies assisting system recovery following disturbance.For example, Allison and Hobbs (2006) used these constructs to understand ongoing degradation on the agricultural lands of Western Australia through a combination of systems thinking and a form of a resilience assessment.They explicated the historical and policy context and the evolving epistemologies of NRM.This was used as the basis for model conceptualization as related to resilience theory and systems dynamics of their case study, and it was followed by a synthesis that included the use of scenarios and an elaboration of the implications for governance and institutions.
Nevertheless, as an emerging field of inquiry, not only were the approaches for resilience analysis still relatively unclear, but the literature did little to shed light on the convergence between various conceptual and analytical frameworks such as systems analysis and resilience assessments or analyses.For example, understanding an issue through the lens of the GAC only partly elucidated for the authors whether the system could be considered resilient and did not help them see what the implications for practice might be.

METHODS: THE DEVELOPMENT OF A SYSTEMIC FRAMEWORK FOR UNDERSTANDING CONTEXT AND CHANGE
As noted, the specific purpose of the initial work on which our discussion is based (Pollard et al. 2008) was to explore the potential impacts of policy changes on water and livelihood security, but to do this in a way that acknowledged the complexities and uncertainties typical of the management of catchment areas.Using the lens of systemic resilience as a means for thinking about such challenges differently, a holistic, systemic framework for understanding context and transformation was developed.This framework was adapted from that of Allison and Hobbs (2006) who combined systems thinking and an early form of a resilience assessment to understand degradation on the agricultural lands of Western Australia (as discussed previously).In our adaptation, the following key changes feature: Ecology and Society 19(2): 63 http://www.ecologyandsociety.org/vol19/iss2/art63/ .An explicit inclusion of learnings from strategic adaptive management (Rogers and Biggs 1999), specifically the V-STEEP process, a heuristic prompting a comprehensive inclusion of factors, namely values, social, technological, ecological, economic, and political. .In the resilience analysis "step," the GAC used by RA and Allison and Hobbs (2006) and certain attributes that confer resilience (Walker andSalt 2006, RA 2007b), plus others added by the authors, were all applied.
. A transdisciplinary, participatory review process by specialists including social scientists, ecologists, climate change specialists, educationalists, agriculturalists, and medical practitioners, the majority of whom had worked in or were familiar with the catchment, was included.This process of specialist involvement was designed to develop and test systems representations of the catchment, to discuss and evaluate thresholds and state changes, and to use these to develop and debate scenarios.The specialists were also asked to reflect on whether the approach generated a systemic understanding of resilience in such catchments and could meaningfully enhance management decisions and practice among a variety of stakeholders and levels.Our key concerns included the opinions of these participants.
Our integrative framework comprised 10 key steps outlined subsequently, the order of which implies a broad sequence, although there were several iterations between a number of steps.In practice, the specialists were engaged informally by the core team, i.e., the authors, from the start and more formally at a collaborative workshop after the authors had synthesized steps 1-6.This engagement allowed for the development of a broadly collective understanding from the earlier steps, as well as being important for the analysis of resilience and the evaluative steps 9 and 10.
1. Literature review: review of existing catchment data, relevant concepts, tools, and their application.
2. Bounding the system of interest: elucidating external and internal drivers considering resilience of what and to what (Carpenter et al. 2001).
3. Development of a GAC: used together with a timeline to identify timescales, i.e., eras, for analysis.

4.
Development of timelines: to elucidate drivers and variables across time and to inform the selection of eras.

5.
Holistic framing of context: description of the sociopolitical, institutional, and environmental context of eras, using the V-STEEP process described subsequently.

6.
Develop systemic view of the SES, i.e., ongoing iterations: development of causal-loop diagrams (CLDs) as descriptions of "the SES system" for the selected eras.
8. Narrate the systemic view: through the consolidation of qualitative and quantitative data, i.e., through iterations of CLDs in which additional data were incorporated.

9.
Qualitative resilience analysis and scenario development: both done collaboratively with the specialist group.
10. Implications and recommendations: consideration of these for management.
Apart from reflecting on experiences from the development and implementation of the previous suite of tools, we also reflect on some examples of the wider societal influence in terms of policy and outcomes that the work appears to have had in the five odd years since the work was initiated or that it may have in the future.These examples were chosen from the authors' own further experiences, and hence the list is not exhaustive and carries qualifiers.

RESULTS: REFLECTIONS ON ADOPTING A SYSTEMIC FRAMEWORK
Table 1 provides a synthesis of each step, its purpose, and the various evaluative inputs of both the authors and specialist group.
The discussion provides more generic reflections regarding the application of our approach by practitioners and researchers in similar contexts.
The notions of resilience and vulnerability were critical components of the framework in terms of understanding policy reform and change.As noted, because a framework for a resilience analysis (RA 2007b) was not yet available when this work began, the team thus used a combination of systems thinking and the Allison and Hobbs (2006) approach together with published criteria known to confer resilience (Walker and Salt 2006), plus three criteria added by the authors, i.e., cross-scale, variability, and nature of learning.Comparison between our approach and that of the subsequently published RA workbooks shows a high degree of congruence.The most important differences relate to (1) high levels of effort in understanding history through timelines; (2) more formal scoping of context using the V-STEEP framework, which explicitly includes societal values; and (3) the greater emphasis placed on a systemic model of the SES through the collaborative development of the CLDs.
Step 1: In this step, studies that described the socio-political and ecological context of the catchment were reviewed as the basis for constructing the systemic view.This is a critical step in that it provides the basis for the systems analysis and a review of the empirical data on which to proceed through to the next steps.Further, a review of the conceptual framing also identified gaps in understanding or the need for greater coherency between different, although related, processes and approaches.
Steps 2 to 4: Although "systems" are models created to support understanding, and hence system boundaries are artificial, the selection of boundaries is designed to best suit the purpose of the work (see Ulrich and Reynolds 2010) so that studies and processes can be bounded as internal, external, or ignored (see Allison and Hobbs 2006).Given the focus on catchment water security, the catchment represented the spatial boundary with land-use zones, i.e., A, B, and C, as described previously, as subdivisions.Initially, the temporal boundaries chosen represented the arrival of http://www.ecologyandsociety.org/vol19/iss2/art63/Table 1.Summary of the broad steps comprising the integrative framework, the purpose of each, and an evaluation of its application in this study.

Broad steps Purpose Evaluation
Step 1. Literature review (catchment, concepts and tools) Deals with variability across disciplines and eras.Starts the process.
An important step that provided the conceptual grounding of multiple approaches and an overview for V-STEEP (acronym for values, social, technological, economic, environmental, and politico-legal).
Step 2. Description of the system boundary Facilitates consideration of "resilience of what, to what" (Carpenter et al. 2001).Bounds data needs and analytical scales.Allows for identification of external and internal drivers.
Catchment was chosen as the system in keeping with policy and the focus of the work.Internal zones proved useful to bound the analysis.Delineation of eras helped focus the analysis.
Step 3. Generalized adaptive cycle (with time line); and panarchy analysis of nested GACs Recognizing that systems go through collapse and renewal, the purpose is to identify such phases and to use these to elucidate windows of opportunity.The panarchy analysis does the same but for a nested hierarchy of GACs (see Holling 2001) and thus aims to identify multiple and cross-scale factors.
Because no methodological guidance is given for their application, the process was largely a subjective "retro-fitting" of historical events.GAC may offer a useful heuristic but given the above constraints they added only marginally to overall analysis.The GAC helped understand some concepts such as brittleness, lags, and opportunities provided by renewal.
Step 4. Development of timelines Helped (a) set the temporal boundaries; (b) elucidate drivers across time and informed the selection of eras for analysis.
Informed the selection of temporal boundaries.Supported the categorization of events as broadly biophysical, political, and social which was important for naming drivers and the causal-loop diagram (CLD) construction.
Step 5. Holistic description of the system using V-STEEP and collection of data for characteristics of resilience Broad but holistic scan of the context (bounded in time), helping to indentify drivers and factors; data sources and availability; Compilation of data pertaining to the characteristics of resilience.
Started the process of developing cause-and-effect linkages and understanding of the catchment as a system.Broaden scope of team understanding beyond specialist comfort zones to focus on water and livelihood security.Useful compilation of data and discussions pertaining to characteristics of resilience although often only qualitative in nature.
Step 6. Development of causal loop diagrams (CLD) Collaborative and visual consensus-seeking process to develop a depiction (model) of the system, and to identify key drivers and interrelationships and feedbacks.
Represented the heart of the work.Strength lies in the coconstruction process especially from different perspectives.Useful to seek joint understanding on feedbacks.The "3-8 rule of hand" is used to identify critical drivers (see Allison and Hobbs 2006).
Step 7: Broad specialist engagement with iterations of steps 4-6; and initiation of collaborative approaches to steps 8-10.This step included a formal evaluation of the each of the steps The purpose was to develop a collaborative (specialist) understanding of the system through comment and discussion; and to initiate steps 8-10.
Strength lies in the coconstruction process especially with participants from different backgrounds/disciplines. Useful to seek joint understanding on feedbacks.Extremely useful for specialist views on resilience and scenarios that were evaluated collaboratively.If unconstrained by participants' time, this process could be undertaken a number of times and should include key stakeholders, e.g., government officials, sector representatives.
Step 8. Consolidate data to populate systems the CLD Data were synthesized using overarching CLD.Linkages in particular were explored.
Reviewing huge amounts of data requires a framework (such as the CLD) for synthesis; Collaborative analysis and synthesis is imperative Step 9. Qualitative resilience analysis and the development of scenarios The key aim of resilience analysis is to identify drivers and system thresholds, their nature, and what determines movement of a system from one system configuration to another, e.g., from desirable to undesirable (Walker et al. 2002).
This study did not have the benefit of the steps outlined in the workbooks.However, when compared later, this framework was broadly parsimonious with that of the Resilience Alliance (RA).The attributes for assessing resilience (Walker and Salt 2006) are a useful set of criteria that can be added to.However there is little guidance on using these criteria in practice to assess resilience.Allison and Hobbs (2006) used only GAC and related cycles for this step; here more emphasis placed on RA criteria which were later expanded through this work.
Step 10.Implications and recommendations for management To find out whether this approach is useful and exactly how to use it for managers, policy makers and in practice.
This process emphasizes the cocreation of knowledge rather than developing credible ex post facto recommendations to an independent decision maker.However many credible and more resilient decisions will be taken if this includes key stakeholders.Having this, a wide diffusion of this type of approach in a variety of natural resource management settings in the region, is underway.http://www.ecologyandsociety.org/vol19/iss2/art63/colonizers in the late 1800s to the present.Subsequently, however, a narrower timescale was used to reflect the core focus of the work, i.e., the impacts of policies on water and livelihood security, namely 1950-1993 and 1994 to the present.Refining the timescale was supported by the construction of the timeline (Appendix 1) together with the first conceptualization of the system using the GAC (see Appendix 2).The timeline is, as Cilliers (2000) points out, an important part of understanding the role of historic events in determining key drivers within a system.This iteration suggested that previous policies, i.e., prior to 1994, developed after the Nationalist Party came to power in 1948 (Appendix 1), gave rise to key political and economic drivers that fundamentally influenced water and livelihood security in the study system.They shaped the creation of separate homelands for people classified as "black" and forced removal into parts of the SRC in the late 1960s, the reduction in quality and scope of education in these homelands, and the need to access cheap labor pools for mining around Johannesburg (Beinart 2001).
This stage of the work included various iterations between the timeline and the GAC.The application of the GAC (see Appendix 2) suggested that the apartheid era had become extremely resilient, meaning that intended outcomes were persistent, but "brittle" in the RA parlance, i.e., little room for new ideas or adaptability to changing external drivers.Following collapse, the post-1994 period could be interpreted as one of renewal.Steps 3 and 4 included the compilation and synthesis of data pertaining to the characteristics of resilience.
Steps 5 to 8: The collaborative construction of CLDs of the SRC as an SES during different eras was assisted by a detailed description of five broad categories of attributes of a system abbreviated as V-STEEP, namely values, social, technological, ecological, economic, and political (Biggs and Rogers 2003; Appendix 3).The V-STEEP description produces a more comprehensive scoping of contextual factors than conventionally examined, which, importantly, assist in identifying a wide spectrum of key drivers.This "SES system" was then represented visually, with inputs from specialists, as CLDs that provided a broadly consensual and systemic graphic model of the drivers, variables, and interlinkages during the era 1950-1993 (Fig. 2) and the postapartheid era after 1994 (Fig. 3).This suggested that economic and political drivers during the apartheid era had been instrumental in shaping the SES system so that water and livelihood insecurity, including weakening social capital and distorted family composition, emerged from persistent reinforcing feedbacks.This analysis further suggested that despite new policies (Fig. 3), inherent lags have meant that these reinforcing feedbacks have persisted.Nonetheless, some changes are emerging such as new institutional arrangements for water and the introduction of child-care grants, which have had a significant impact on livelihood vulnerability (DSD et al. 2012).
Following this, the group examined qualitative and quantitative data regarding the proposed characteristics of resilience such as diversity, ecological variability, the role of slow variables, polycentric governance, social capital, the breadth of ecosystems services, innovation, and openness.With little guidance from the literature on how to use these to assess resilience, they were finally ranked as low, moderate, and high under each of the scenarios examined.In terms of the current scenario, most of these were either moderate or high, except for feedbacks, social capital, and innovation, all of which were low.In contrast to commonly held perceptions, most specialists agreed that although degraded, the terrestrial ecological system had not shifted states.In contrast, conclusions regarding regime shifts were less clear in terms of the riverine ecosystem of the Sand River, which has experienced some flow cessations in dry years, although recent work points to a high degree of resilience of the riparian vegetation in Zone B (K. Kotschy, personal communication), highlighting the multidimensional nature of change.However, the consensual view was that social states had shifted.With the migration of adult males to the mines in the 1970s and 1980s, the extended patriarchal family, and even the nuclear family unit, appears to have been replaced by a more dispersed sibling social network (Niehaus 2002;I. Niehaus, personal communication).These social networks have been further challenged by ongoing shocks such as high HIV/AIDS infection rates (Hunter 2007, Kahn et al. 2007).Steps 9 to 10: The CLDs and preceding information were used to formulate, collaboratively, three scenarios relevant to water governance.This approach contributed to the resilience analysis through emphasizing trends in space and time; issues of polycentric governance; learning and leadership in the catchment; decisions regarding key fast and slow variables, which were operative at different scales; and the major reinforcing and counterbalancing feedbacks.Finally, the group reflected on whether, and how, this approach might be useful for management and policy analysis (see Table 1).
To conclude, Table 2 summarizes the authors' reflections on the potential influence that such thinking has had on wider NRM issues by looking at the longer term outcomes.Although it is always difficult to determine when these effects can be deemed to be the specific result of this original work, as opposed, for instance, to the general spread of these ideas more broadly in society, the cases shown in Table 2 are very closely aligned to persons and specific ideas from this original work, suggesting a strong influence.The results featured in Table 2 in no way suggest that the thinking discussed by us has been singularly or even predominantly responsible for all the broader changes listed, but rather that this approach and decision-making style has been influential in the transformation of both policy and practice in some way.Time will tell how significant this is.

DISCUSSION
The combined suite of steps based on systemic approaches described previously provided a fresh approach not only to assessing systemic degradation and resilience, but also for understanding the impacts of policy changes in shaping water security and livelihoods in the SRC.The exercise pointed to a tightly coupled, resilient system during the apartheid era, which despite major policy changes, will take time to transform to one that gives meaning to the principles of equity and sustainability espoused in the postapartheid policy changes.
Nonetheless, those involved in the reflections, i.e., the authors on an ongoing basis and the specialists at the collaborative workshop and informal interactions, felt that some steps proved more useful than others, with boundary setting, a holistic understanding of context through timelines, and the V-STEEP scoping, a collaboratively derived systems view through CLDs and the resilience analysis, being the most valuable.To a lesser extent, scenario generation proved useful to some.Although the first steps of conceptual and literature reviews are obvious prerequisites for subsequent steps, our experience was that it took a while to convince all participants of the need to invest effort in defining the system boundaries or to understand history and context as thoroughly as intended in the framework.We assert that it was only by constraining the system through spatial and temporal boundaries, and through carefully inter-relating historical elements, that a credible systemic description emerged.Moreover, deriving as holistic a contextual understanding as possible greatly facilitated the subsequent transdisciplinary interactions and helped identify multiple cross-scale drivers.The discussion of values, i.e., deeply held beliefs, identified by the V-STEEP heuristic proved an important exploratory area with the specialists.The use of the GAC, although offering a useful heuristic to consider collapse and renewal, proved difficult to use at multiple scales as part of a resilience analysis and added only marginally to an overall understanding.
The contribution of systemic representations such as CLDs to interdisciplinary interaction was central because it is in the process of coconstruction that a shared integrative view becomes possible.The process also highlighted key areas of contestation.An advantage of such system-wide depiction is that social, economic, biophysical, and other drivers are captured in a visual, systemic view facilitating dialogue between participants.The systemic representation of the catchment over two eras facilitated the identification of key drivers, interlinkages, and feedbacks that may otherwise have been overlooked.For example, the link between livelihoods and the migration of adult males (see Fig. 2) helped participants think about history and to reflect on the notion that the strongest drivers over the past hundred years were indeed political and economic in nature, and cross-scale in their effects.Clear acknowledgement of the feedbacks in the two persistent reinforcing loops, i.e., ecosystem and social, in Figure 2 prompted important discussions.Many felt that unlike the "lock-in" traps described by Allison and Hobbs (2006), these persisted because of lags, rather than being permanently "locked in," so that when the CMAs are operational, these may well change.
Additionally, the systemic representation highlighted the nuances of mediating social-ecological interactions.For example, instead of opting for simplistic solutions, such as "government must regulate," when considering the illegal use of various natural resources, participants could appreciate the complex nature of interactions and cross-scale influences that had led to this situation.Poverty, weakening local-level institutional arrangements together with the lack of governmental capacity to act, and the uncertainties rendered by land reform have all meant that natural resources are increasingly vulnerable to opportunistic interests (Cousins 2007, Pollard andDu Toit 2011b).This suggested that without strong local-level and polycentric governance systems, a key factor thought to confer resilience, the sustainability of systems is likely to be compromised (Ostrom and Cox 2010).
The involvement of a transdisciplinary group of specialists proved invaluable, but upon reflection, the authors felt that this group should have been more broadly selected, including policy makers and a wider range of practitioners, and more intensely involved from the start.In addition to the aforementioned benefits of collaboration, the varied group was able to contribute to ideas on factors that might confer resilience in addition to those that are conventionally recognized by the RA.For instance, much discussion focused on the nature of learning within SESs where ultimately it was suggested that the content of the learning is far less important than the nature of the learning process.Social learning approaches suggest that there is a greater likelihood of sustainability emerging within a particular context through learning that is based on reflexive and adaptive social processes (Ison et al. 2007, Pollard and Du Toit 2007, Wals 2007), and exercises like these as well as "interventions" need to be attentive to this.
The several years that have elapsed since this work have provided the opportunity to trace examples of the medium-term and perhaps longer term influence of the thinking that was developed in this work (Table 2).For example, the framing of the South African catchment management strategy guidelines is almost wholly based on the paradigms used in this work.Several key Water Research Commission projects build on this foundation, Table 2. Recent examples of how the thinking developed in this paper has influenced, or is potentially likely to influence, policy change and implementation regarding water and livelihood security in catchments in South Africa.The work was initially described in a detailed report (Pollard et al. 2008) by which time some of the initiatives below had already begun.Many other factors and influences often also contributed to these changes.Since 1995 but systemic approaches more particularly since 2005 after visit of Helen Allison to Kruger National Park.There has over this whole period been close coevolution of concepts and thinking between those described in this paper and the Kruger National Park, which also has a special interest in rivers crossing the park, including the Sand River.For instance, SANParks now uses CLDs (local use pioneered mostly by authors of this apper) in most of its Science-Management in 20 national parks across the country.
Because of the concern around rivers in National Parks, and because of a key link between persons working in and outside parks in the same catchment, it is possible that this collaboration will continue.e.g., studies of collaborative approaches to water-related ecosystems services, and large current development-oriented projects aimed at adaptation to climate change, e.g., Resilience in the Limpopo Basin, or RESILIM, hold these concepts at their core.Although there are many other significant influences, it is thus justifiable to suggest that policy changes in South Africa regarding water and related livelihood security have been materially influenced by this approach, which has thus enhanced understanding in the way intended.

CONCLUSION
The unique contribution of this systemic approach is that it engages with SES issues such as water and livelihood security in a manner that acknowledges the importance of (1) systems and resilience thinking and (2) consensus seeking through approaches that enable coconstruction of the processes of engagement and problem framing.This approach is similar in many respects to the mainstream RA workbook approach but differs in the emphasis it places on deriving a systemic view and on the wide scan of drivers elaborated through the V-STEEP process, as well as the moderately greater emphasis on history through explicit timelines.
The general combination of systems thinking and resilience analysis approaches facilitated the development of an integrative, systemic understanding of change and transformation at a catchment scale by a diverse group of specialists who, to some extent, represented multiple groups of stakeholders.These specialists agreed generally that the framework elucidated key http://www.ecologyandsociety.org/vol19/iss2/art63/drivers, inter-relationships, and feedbacks in the system; highlighted data needs; and was useful for "seeing managerial issues" from a broader systemic perspective.
On the other hand, there was ongoing frustration on the part of some at not having "sufficient" empirical data to develop a "quantified systems view."Although they felt that comprehensive empirical data were critical for a "full" understanding of the system to manage it, the fact that our work was better endowed with data sets than most in the country indicates that this is extremely unlikely in practice: catchment managers have to act in a world of uncertainty and "incomplete" information.This is not to suggest that systematic and empirical inquiry is not critical.Rather, the realization is that we face a world of uncertainty and rapid transformation that also requires dealing with the unknown, highlighting ongoing learning and adaptation.In catchments, everything cannot be known or verified with empirical data (Heylighen et al. 2007), even as we promote as many empirical studies as practicable.Catchments, therefore, constitute contexts that lend themselves to the application of systems thinking as a core approach, not as one that is simply tagged onto a systematic or evidence-based drive.We suggest that multiple groups of stakeholders in the arena of IWRM are indeed starting, in various ways, to use this approach, and that policy-influential documents and programs are adopting these.
Finally, depictions of systems such as those presented by us should not be conflated with attempts to depict the truth.Rather, they are better seen as models or heuristics of what is known about the system, which, through participatory processes of representation and narrative, can suggest potential constraints, bottlenecks, and feedbacks.In turn, these enrich managerial responses and stakeholder dialogue.As Cilliers (2001:3) points out, "This is not because of some inadequacy in our modeling techniques, but a result of the meaning of the notions 'model' and 'complex'.There will always be a gap between the two.This gap should serve as a creative impulse that continually challenges us to transform our models, not as a reason to give up." Responses to this article can be read online at: http://www.ecologyandsociety.org/issues/responses.php/6312

Fig. 2 .
Fig. 2. Causal-loop diagram exploring a systemic view of water security in the Sand River Catchment social-ecological system for the era 1950-1993.

Fig. 3 .
Fig. 3. Causal-loop diagram exploring a systemic view of water security in the Sand River Catchment social-ecological system for the postapartheid era (after 1994).
The Wise Use project was initiated in 2009 in an attempt to build local custodianship over wetlands being rehabilitated through the Working for Wetlands Programme.The Wise Use approach is founded on a systemic, social learning approach that supports local land owners or users to develop a systemic view of degradation and sustainability.The approach is being developed and tested at learning sites.