Barriers to mainstream adoption of catchment wide Natural Flood Management, a transdisciplinary problem framing study of delivery practice

Natural Flood Management (NFM) is the name given to Nature Based Solutions (NBS) for Flood Management in the UK. It is a holistic flood management technique that employs natural hydrological processes, through the instillation of 15 interventions, to slow the flow of water, creating a landscape scale flood management system. Despite widespread interest and supporting policy from governments and non-profit organisations NFM, as yet, has not been widely adopted as a mainstream flood management technique. A small number of academic studies examining perceived barriers to NFM adoption have identified a variety of individual factors as being responsible. It is commonly accepted that flood risk management broadly, and NFM specifically, are complex, challenges of interacting physical and human parameters and that academic, institutional 20 and policy divisions are rarely sympathetic to embracing these complexities. A transdisciplinary problem framing study in conjunction with professionals experienced in the delivery of NFM projects in the UK aimed to capture these multifaceted parameters of flood management and strategic delivery at a landscape scale using Group Concept Mapping, a systems approach to identify conceptual convergence. This policy-delivery impasse was further explored by quantifying the relative importance of individual barriers and conceptual groupings from the perspective of two different practitioner groups (flood risk managers 25 and conservation practitioners). The results demonstrate that the NFM delivery system can be grouped into seven interacting elements: policy and regulation, politics, public perception, cross-cutting issues, funding, technical knowledge and evidence, of which each have a varying number of barriers that limit NFM uptake. Opinions differs as to the importance of these individual barriers, however when considering the system broadly we identify that the institutional and social barriers are perceived as the most important, whilst technical knowledge and evidence are the areas of least concern. This paper aims to 30 promote NBS flood management delivery in the UK and globally by generating, structuring and representing the multifaceted and multilevel NFM delivery system at a local level to evidence adaptive decision making at a regional, national and global https://doi.org/10.5194/hess-2021-404 Preprint. Discussion started: 30 July 2021 c © Author(s) 2021. CC BY 4.0 License.

Abstract. Natural flood management (NFM) is the name given to nature-based solutions (NBS) for flood management in the UK. It is a holistic flood management technique that employs natural hydrological processes, through the installation of interventions, to slow the flow of water, creating a landscape-scale flood management system. Despite widespread interest and supporting policy from governments and non-profit organisations, NFM, as yet, has not been widely adopted as a mainstream flood management technique. A small number of academic studies examining perceived barriers to NFM adoption have identified a variety of individual factors as being responsible. It is commonly accepted that flood risk management broadly, and NFM specifically, are complex, challenges of interacting physical and human parameters, and that academic, institutional and policy divisions are rarely sympathetic to embracing these complexities. A transdisciplinary problem-framing study in conjunction with professionals experienced in the delivery of NFM projects in the UK aimed to capture these multifaceted parameters of flood management and strategic delivery at a landscape scale using group concept mapping, a systems approach to identify conceptual convergence. This policy-delivery impasse was further explored by quantifying the relative importance of individual barriers and conceptual groupings from the perspective of two different practitioner groups (flood risk managers and conservation practitioners). The results demonstrate that the NFM delivery system can be grouped into seven interacting elements, policy and regulation, politics, public perception, cross-cutting issues, funding, technical knowledge and evidence, of which each has a varying number of barriers that limit NFM uptake. Opinions differ as to the importance of these individual barriers; however, when considering the system broadly we identify that the institutional and social barriers are perceived as the most important, whilst technical knowledge and evidence are the areas of least concern. This paper aims to promote NBS flood management delivery in the UK and globally by generating, structuring and representing the multifaceted and multilevel NFM delivery system at a local level to evidence adaptive decision making at regional, national and global levels. Through problem structuring and an increased understanding and awareness of the structure and network of linking elements and perceived differences of practitioner groups that influence the system of delivery, steps can be taken towards solutions that are socially, scientifically and practically robust.

Introduction
In the UK natural flood management (NFM) is the commonly used term for nature-based solutions (NBS) for flood management, a holistic flood management technique designed to mimic natural environmental conditions by harnessing hydrological processes to slow water flowing through the landscape (Werritty, 2006). The terms NFM and NBS for flood management (WWAP, 2018) indicate flood management as the primary goal of the techniques. In practice, proponents of their use emphasise their transformative strength in the delivery of multiple benefits (Barlow et al., 2014;Forbes et al., 2015;Hanson et al., 2020). Whilst some such as Dadson et al. (2017) have raised concerns that during extreme floods measures could be overwhelmed, others such as Norbury et al. (2021) have observed that this may not necessarily be the case, subject to design and magnitude of event. Rather, the paradigm shift for flood management in adopting NBS within their programmes is a contribution to flood risk reduction alongside greater environmental and social goods (Connelly et al., 2020;Fenner, 2017;Hanson et al., 2020), including improved biodiversity (Cook et al., 2016), water quality (Barber and Quinn, 2012;Howe and White, 2003), public health and well-being (de Bell et al., 2017;Bratman et al., 2019;Maas, 2006) and carbon sequestration (WWAP, 2018). NFM describes methods that restore (Lane, 2017) or mimic (Barber and Quinn, 2012) hydrological processes within the water cycle, including engineered land forms (Wingfield et al., 2019). The pinnacle of the practice is to link a large number of small interventions through connectivity (Keesstra et al., 2018) to optimise a system functioning at a landscape scale delivering a cumulative catchment-wide land and water management strategy. Sectors needed to deliver coordinated activities, at multiple scales, for strategic delivery include urban planning and development, agriculture and conservation, flood and coastal risk management and water and wastewater supply and management. At present the activities and legal frameworks of these sectors in the UK are not aligned (Goytia, 2021), nor does NFM have a well-defined delivery programme with a clear champion to drive real action and/or empower others to change (Wingfield et al., 2019). To date NFM has not been widely adopted as a mainstream flood management technique (Bark et al., 2021), with some practitioners continuing to regard it as a novel approach (O'Donnell et al., 2017;Schanze, 2017).
Research is required to reveal transformative pathways for the incorporation of NFM into a mainstream flood management strategy that conceptualises NFM as dynamically complex, which can reveal associations between social and environmental domains (Gómez Martín et al., 2020;Keesstra et al., 2018). Within social-ecological systems multiple actors bring differing perspectives of the nature of problem(s), different interpretations of potential solutions and ambiguity as to who is responsible for applying and operating any new method. Academic disciplines, and the career and journal systems that serve them, generate epistemological limitations creating barriers between actors in problem framing, application of methods and ideals of "proof" (Hazard et al., 2020;Mauser et al., 2013). Furthermore, a technocratic model of transmitting scientific fact without engaging in dialogue across disciplines and between science and society is seen in the academic literature's focus on NFM effectiveness as a measure of flood peak reduction or delay (Connelly et al., 2020;Dadson et al., 2017;Wingfield et al., 2019), which is insufficient for transformative decision making (Young et al., 2014). For example, NBS for flood management require land and, as such, land ownership, motivations of landowners (Slavíková and Raška, 2019), balancing of public and private interests and clarity on whether interconnected policy, legal and economic systems are supporting or hindering the engagement of landowners (Hartmann et al., 2019), and therefore the commitment of land is critical to mainstreaming NBS. A growing number of voices concerned with sustainable futures have called for a change to scientific knowledge generation to instead foster knowledge co-production with academic and non-academic actors to develop integrated research questions, services, policies and processes (Mauser et al., 2013;Pereira et al., 2018;Schneider et al., 2021). System-oriented research, also known as interdisciplinary or transdisciplinary research, aims to "transcend disciplinary and methodological paradigms" (Hadorn et al., 2008) to foster knowledge exchange and generate new mental models and practices for sustainability-oriented action (Pereira et al., 2018).
Environmental practitioners work within an established system reinforced by embedded ways of working, governing and thinking about the system (Bark et al., 2021;Ngai et al., 2020;O'Donnell et al., 2018). The primary focus of NFM guidance written for practitioners is on supporting quantification of ecosystem processes and engineering principles to design interventions (Burgess-Gamble et al., 2017;CIRIA, 2018;NWRM, 2013;Environment Agency, 2017). Guidance documents are not explicitly addressing the interdisciplinary nature of NFM research, planning and delivery. Mainstream delivery of NFM requires a paradigm shift that incorporates social, environmental and ecological dimensions (Dekker and Fantini, 2020;Werritty, 2006); those practitioners with responsibilities for designing, applying for funding and/or delivering projects have valuable knowledge and experiences from navigating these complex dynamics.
This paper describes a methodological approach and outputs of a problem-framing component of a wider transdisciplinary study (2016)(2017)(2018)(2019)(2020) between academics and practitioners in the UK with a shared interest in promoting the mainstream adoption of NFM. The authors are a doctoral research student who designed and led the project and a multidisciplinary and practitioner supervisory team that draws interdisciplinary expertise from flood research, human geography, governance, time space and territory, catchment science, fund raising and partnership working. The process of a transdisciplinary research programme begins with problem framing, knowledge co-production and integration of interested actors (Norström et al., 2020;Pohl et al., 2021;Schneider et al., 2021). To do this, group concept mapping (GCM) (Trochim, 1989) with Ketso (Tippett et al., 2007), a systems thinking methodology (Hassmiller Lich et al., 2017) that applies a mixed-method, data collection approach was used. While not explicitly developed for use in transdisciplinary research, it is a useful tool, allowing for the emergence of contested knowledge, differences in knowledge framing or highlighting blockages in knowledge transfer. The GCM method produces visual representations of what a group is thinking on a particular topic (Donnelly and Ph, 2016) and in doing so enables integrated problem identification, the primary component of transdisciplinary research (Jahn et al., 2012;Lang et al., 2012;Pohl et al., 2021).
Rather than limit the study to merely identifying individual barriers to NFM delivery, our aim is to examine interdependencies and identify conceptual convergence within the delivery system. In doing so the study reveals conditions under which barriers to the delivery of NFM persist and begins to identify areas for further research and intervention points that could act as a catalyst for change (Eisenack et al., 2014). Our aim in describing the methodological approach and sharing findings is to support NBS adaptive decision making in contexts outside of north-western England by revealing and examining problems of this particular system . In-depth case studies, when used and reviewed alongside the natural science evidence base, can act as a cata-lyst to "activate" hydrology research (Maruyama, 2001). The principle of connecting research practice and theory with input from practitioners guided our approach, and in doing so the outputs have the potential to be transferred to other settings (Bickman et al., 2016). It is recognised that while data collection is limited to England, questions on how to implement NBS for flood management are of interest globally (Gómez Martín et al., 2020;Sowińska-Świerkosz and García, 2021). Contexts and actors involved differ across the globe, but our systematic method that brings together diverse groups of actors in a process of knowledge co-production is informative globally. Our approach can be replicated by multidisciplinary groups interested in comparing, contrasting and evaluating patterns and structures within environmental management systems to identify transformative pathways and knowledge gaps.
The paper begins by demonstrating a discrepancy between policy and delivery in relation to landscape-scale holistic flood management techniques and frames NFM delivery as a wicked problem requiring a research approach that transcends disciplinary and methodological paradigms integrating knowledge of academic and non-academic actors. The paper then proceeds to investigate how group concept mapping with Ketso was used as a structured method to engage in dialogue and co-create knowledge across disciplines and practitioner groups. The paper first explains an established approach of studying barriers and adopting sustainable management paradigms and then combines this approach with transdisciplinary problem framing and a step-by-step reflection on using group concept mapping with Ketso in practice. It does so to demonstrate its potential and provide a framework for integrative systematic problem-oriented research that creates visual representations of networks of ideas to describe how a group thinks about an issue to be used in developing mutual understanding as a basis for further research and cooperation.

Barriers
Barriers are widely studied in reference to modern sustainable management paradigms, such as climate change adaptation Eisenack et al., 2014;Moser and Ekstrom, 2010) and sustainable agricultural practices (Bustamante et al., 2014;Rodriguez et al., 2009). Barriers are distinct from limiting factors in that barriers within a system, once they are identified, can be managed and overcome, whereas a limiting factor within that system is insurmountable (Eisenack et al., 2014). Policy documents of environmental regulatory authorities in England encourage the use of NFM, describing it as "an option more resilient to extreme events" (Environment Agency, 2011) and "better value for money and a flexible and resilient solution that can deliver socioeconomic and environmental benefits alongside flood risk management benefits" (Penning-Rowsell, 2010). How-ever, NFM as yet is not embedded as a mainstream flood management strategy (Bark et al., 2021), our understanding on why remains somewhat fragmented and a number of explanations have been put forward (Table 1).
The literature summarised in Table 1 demonstrates that investigations examining NFM delivery are focused principally on the flood risk management sector, with only two of the studies bringing together perspectives of different stakeholder groups. Deficiencies in the evidence base are a theme common amongst papers from 2017, but from 2019 a shift towards mechanisms of delivery can be seen (Environment Agency, 2019; Wingfield et al., 2019). Waylen et al. (2017), while highlighting a number of barriers, including a lack of resources and a need for more evidence, strongly advocate a need for culture change within flood risk management itself. This finding is echoed in the internal review undertaken by the Environment Agency (2019). Institutions responsible for flood management are resistant to change (Buuren et al., 2015;Sarabi et al., 2019); however, NFM delivery by its very nature transforms the scale at which flooding and flood generation processes are considered, from local to catchment. It is largely because of this change in scale that strategic planning and delivery are needed amongst sectors that are not currently coordinating activities (Wingfield et al., 2019). To date no study has been undertaken at this scale, specifically designed to examine delivery practitioner interactions within which NFM delivery is situated, in a process of knowledge co-creation, across disciplines and with academic and nonacademic actors. This study aims to address this gap by targeting practitioners who contribute to catchment partnerships (Collins et al., 2020) and flood and coastal erosion risk management professionals.

Research approach and methodology
NFM delivery is influenced by social, technical and biophysical factors that interact and feed back in a multitude of complex ways, a characteristic commonly described as a "wicked problem" (Lane, 2017;Rittel and Webber, 1973). Unlike "good" problems, which operate within defined rules (Brown et al., 2010), wicked problems are dynamic, and an approach limited to the methods and perspective of one disciplinary domain are insufficient. Research aimed at problem resolution, such as how to deliver catchment-wide NFM, is required to address multifaceted socio-ecological problems involving interdisciplinary research techniques and researchers to work with practitioners to understand and improve the system (Mctaggart, 1991).
Benefits of researchers and practitioners working together are well established in the participatory research literature; indeed, this structure is reflected in the interdisciplinary project team, with one of the authors a director of a host organisation in north-western England. The inclusion of a host organisation director as part of the project team (and as a PhD supervisor) offered insights and provided access to a complex system but did not determine any research priorities nor shaped or determined research findings, as they were not present during workshops or data collection/analysis events. Throughout we were conscious of the positionality of the research team and potential implications for the research process and sought to minimise potential bias and influence, which was an additional benefit of the methodological approach applied. Novel findings and opportunity for change are generated by empowering those individuals who are most familiar with the system under scrutiny to steer the research agenda, data generation and application of findings (Newing et al., 2011). An interdisciplinary, participatory approach embraces socio-ecological complexity (Jantsch, 1972) rather than a traditional position, which tends to smooth out variation. Practitioners of NFM are a heterogeneous group: individuals will differ in their perspectives, ideas and interactions with other actors in response to different disciplinary backgrounds and a range of values and conceptual framings. Research that harnesses processes of knowledge co-production (Mauser et al., 2013;Pereira et al., 2018;Schneider et al., 2021) and social learning to explore and clarify differences can reveal a range of values and framings (Keen et al., 2005;Pahl-Wostl et al., 2007) whilst encouraging deconstruction of ineffectual old ideas and constructing shared new possibilities. Conceptual changes seed shifts in scientific and policy discourse that have been shown to bring about paradigm shifts (Bark et al., 2021;Pahl-Wostl et al., 2011).

Participant identification
The integration of land and water management to harness hydrological processes to increase the hydrological resilience of the system is a messy real-world problem that has failed to move from a 2-decade-old policy ambition. A previous study identified that, within the UK environmental governance system, catchment partnerships are well placed to coordinate delivery as the integrated water management framework that steers the movement is comparable and compatible with NFM in encouraging the delivery of multiple benefits coordinated at a catchment scale (Wingfield et al., 2019). Schanze (2017) gives an alternative view calling for research that advances knowledge to support NFM delivery within flood risk management practice. This is the first study that brings together these two practitioner groups to examine these two perspectives in transdisciplinary and knowledge co-production research (Norström et al., 2020;Pohl et al., 2021;Schneider et al., 2021).

Flood risk authorities
In England, risk assessment management authorities (RMAs), e.g. the Environment Agency, lead local flood authorities (LLFAs), district and borough councils, coast protection authorities, water and sewerage companies, internal 2. Private landowners with "rights" over land use decisions drainage boards and highways authorities, are responsible for the delivery of flood risk management policy. In 2020 the English government published a new national flood and coastal erosion risk management strategy that places greater emphasis, than previous strategies, on nature-based solutions and using catchment-based approaches (DEFRA, 2020). Current flood models do not satisfactorily simulate hydrological processes of an NFM-integrated system at a catchment scale (Metcalfe et al., 2017;Pattison and Lane, 2012); therefore, it can be challenging to demonstrate the "economic" value NFM delivers of flood risk reduction to properties and infrastructure. Furthermore, costs of commissioning modelling studies to justify spending on an NFM scheme can exceed the costs of its delivery (Burgess and Hill, 2018), reinforcing spending skewed towards traditional, hard-engineered flood defence system(s). Flood risk management responsibility in England is divided amongst RMAs according to sources of flooding: river, coastal, groundwater, surface water, reservoir and sewerage. NFM and its use of hydrological processes to slow water in the landscape as a flood risk management technique employed to manage all types of flooding do not align with the delivery responsibilities of one RMA over another.

Catchment partnerships
Catchment partnerships were established in 2011 to deliver integrated water management by bringing together different sectors and organisations into a cooperative forum to facilitate greater integrated land and water management activities. They are led by a "host" organisation, a charitable body, typically an environmental non-government organisation (Collins et al., 2020) working at the local level (Department for Environment Food and Rural Affairs, 2015). Alongside awareness raising, typical of conservation organisations (Mace, 2014), the biological disciplinary framing of host organisations is evident in the majority of their activities designed to enhance biodiversity, protect wildlife and restore habitats (CaBA, 2018). Partnerships use a ground-up model (Koontz and Newig, 2014) to encourage interested parties to collaborate, identify mutual gains and deliver multiple benefits rather than relying on a top-down regulatory approach to drive environmental improvement works. This has become known as the catchment-based approach (CaBA, 2018). It is noted that the CaBA model is not yet proven (Watson, 2015): it faces acute financial uncertainties, receiving limited centralised government funding, and its formation is based on good will rather than regulatory reform. Nevertheless, a review of priority working areas of the 102 different catchment partnerships in England identified flood risk as of interest for 93 % of catchment partnerships (CaBA, 2017) and, given that NFM is an integrated catchment management technique, catchment partnerships are well positioned to lead and influence NFM delivery (Collins et al., 2020).

Group concept mapping
GCM (Trochim, 1989) with Ketso (Tippett et al., 2007) is a mixed-method, data collection approach that visually represents the ideas of a group using multivariate quantitative analysis and creates opportunities for qualitative data interrogation.
The output of GCM with Ketso is described as a map: it displays previously unconnected ideas clustered into concepts that reveal how and what a group and sub-groups think about a topic (Donnelly and Ph, 2016). Crucially, given the participatory nature of this study, it offers an opportunity for the participants to discuss and explore together the interconnected physical and human parameters of the system, incorporating differing perspectives of actors and the significance of diverse or fragmented conceptualisations (Cabrera, 2006;Tippett et al., 2007). Additionally, a quantitative element of the approach allows for an empirical assessment of the perceptions and conceptualisations of a group, resulting in more robust data than achieved through expert opinion alone.
GCM with Ketso consists of six steps, which for the purposes of this transdisciplinary problem-framing study were grouped into three phases: phase-1 qualitative statement generation, phase-2 quantitative statement sorting and ranking and phase-3 interpretation (Fig. 1). Phases 1 and 2 were conducted between April 2016 and September 2016.

Phase-1 qualitative statement generation through Ketso
Phase 1 was completed in two sessions to capture the input from a broad group of water-focused practitioners, the first through a workshop at the River Restoration Conference in Blackpool on 27 April 2016. Thirty-nine of the conference attendees took part in the workshop, enabling access to a heterogeneous practitioner group drawn from across the UK, many of whom are experts in their fields, including contractors, engineers, consultants, academics, environmental nongovernmental organisations (NGOs) and government agencies (Wingfield, 2016). The second workshop was attended by 12 practitioners from the Environment Agency National Capital Programme Management Service (NCPMS), individuals responsible for delivering the Flood Risk Capital Programme. This team was selected as having the most comprehensive knowledge of flood risk management scheme delivery via their responsibility for managing the multi-millionpound budget allocated to large flood management schemes across England.
In both the large (n39) and small (n12) workshops, participants were divided into groups of between six and eight in order to generate statements for phase 2 (Fig. 1). The generation of statements creates a set of ideas that together capture all elements used to describe a system, values and/or concepts of the study domain. A number of techniques can be employed in a research setting to generate statements in- cluding use of a predetermined statement set, text abstraction and keywords; however, brainstorming is perhaps the most commonly used and familiar technique. Brainstorming is most suited to the combination of ideas and knowledge across disciplinary and operational divides as participants express themselves in their own words. This study used a form of live structured brainstorming called Ketso (Tippett and How, 2011;McIntosh and Cockburn-Wootten, 2016). Each group undertook a structured conversation using the Ketso methodology (Tippett and How, 2011;McIntosh and Cockburn-Wootten, 2016), a technique for structuring dialogue in a workshop environment. The advantage of Ketso is that it is specifically designed to allow all participants to contribute without one or a few voices dominating, thereby mirroring the approach of GCM in representing the ideas of a group rather than expert opinion. Second, the technique employs visual, active and oral aids to support the developing discussion (Fig. 2). The Ketso methodology can draw out ideas or themes and make connections that might not otherwise be identified using a traditional focus group discussion (Tippett and How, 2011) or brainstorming activity. A conversation builds in stages around a central question, in this case "how can the delivery of catchment-wide NFM be encouraged?" The first element considered by each group was "what are the foundations of the NFM delivery system?" The second asked "what are the emerging mechanisms that encourage wider NFM delivery?", for example actions, tools, objectives or organisations. The third generated question which forms the basis of this paper is "what are the barriers to delivery?", and finally the last was "what could be done to encourage and develop the emerging mechanisms or address any of the barriers?" A formal definition of a barrier to NFM was not made for the study. The participants were left to interpret this for themselves to avoid consciously or sub-consciously influencing their understanding and therefore potentially their view of challenges, restrictions or problems of NFM delivery. This subjectivity becomes data in themselves and is explored in the second research question as to the differences in perceptions between practitioners.

Phase-2 qualitative statement sorting and ranking
A total of 253 barrier statements were generated during phase 1; these were sorted and reduced to a final list of 47 for the second phase of the group concept mapping methodology. The process had three purposes: 1. reduce the number of statements to a reasonable number for participants to manage in one session whilst ensuring each barrier remains distinct to make comparisons; T. Wingfield et al.: Barriers to mainstream adoption of catchment-wide natural flood management 2. check each statement for clarity and remove any deemed incomprehensible; and 3. combine similar statements (Cabrera, 2006).
An independent reviewer was used to assist the primary researcher and review the validity of the initial statement reduction. Phase 2 was not undertaken in a workshop setting; qualitative statement sorting and ranking (phase 2, Fig. 1) were undertaken by 12 flood risk management professionals and 12 practitioners who contribute to catchment partnership either alone or in a small number of cases in pairs and groups of three. The principal researcher provided guidance to the participants. Whilst national experts were the target of phase 1, statement generation, practitioners who work within north-western England were selected for the second phasestatement sorting and ranking. We were interested in testing the experiences and perceptions of the two different professional groups resulting from planning and delivering NFM interventions and therefore wanted to reduce any influence of biophysical and socio-technical factors resulting from practices in different geographical regions that may complicate comparisons (Dupuis and Biesbroek, 2013). This two-stage process helps reduce the significance of local place or site within the study; however, the national legislative context remains important in framing policy. Phase 2 involved each participant sorting the statements into groups that they felt contained similar or related ideas and classifying the group by giving it a name. Participants were then asked to rank the importance of each individual statement using a 0-7 Likert scale. The scale was low to high, 1 indicating that the barrier is unimportant or disagreed with and 7 indicating that the barrier is extremely important; 0 was included to enable practitioners who perceived they did not have sufficient knowledge to give an opinion.
All 24 participants sort and rank the statements (n47), after which these qualitative data are transformed through quantitative analysis (Step 4, Fig. 1) to reveal the way in which the group as a whole and the two practitioner subgroups create meaning. A group similarity matrix (Trochim, 1989) combines all 24 participants' outputs into a grid square with as many columns and rows as statements (n47), from which a concept map is drawn using multidimensional scaling (Hout et al., 2013;Kruskal and Wish, 1978). Each statement is located on a grid square according to their similarity and dissimilarity. For this analysis a two-dimensional solution locates points representing each barrier statement on an X-Y graph (Fig. 3a). Ward's cluster analysis is applied to the multidimensional scaling outputs. Clustering statements into concept groups allows for many individual ideas that could otherwise be overwhelmed to be brought together via a "higher order meaning-making device" (Goldman and . A level of relatedness is inferred by the proximity of both the individual statements to each other and the proximity of the clusters to each other. Those located at a distance from each other are conceptualised by the participants as being less similar than those that are neighbours. Discretion and knowledge of the system are used by the analysts to propose the final number of clusters, each of which represents a concept within the system. The first output of the cluster analysis maps every single statement as an individual concept, in this case 47 individual concepts. The method then combines statements in turn according to similarity. In the first instance the two statements that were most frequently combined by participants in phase 2 (Fig. 1) form the first cluster. The analysis continues stepwise, combining individual statements into groups (cluster groupings) of statements until the final solution places all statements in a single cluster. The challenge is then to decide which arrangement of cluster groupings best represents the concepts and system (Kane and Trochim, 2007). The final solution depends on the context in which the map is to be used and the level of detail required. It is helpful to sort in two stages, beginning with an initial analysis that follows Ward's cluster analysis through to its completion of one single cluster. Following this a second detailed review begins with agreeing on a minimum and maximum number of concepts (Fig. 3b) to represent the system. Beginning with the minimum number of clusters, the statements or clusters that combine at each step are examined closely, looking for a sense of cohesion, mutual understanding and participant interpretation within the grouping and a difference between the two newly formed concepts. This process is repeated until the clusters are recognised by the participants as concepts that represent the system under scrutiny.

Phase-3 participant interpretation and analysis
The final step of the method brings participants together to interrogate and reflect on the analysis and interpretation of the previous two phases (Bickman et al., 2016;Newing et al., 2011). On 30 April 2018 in Warrington, north-western England, 19 practitioners attended a workshop to contribute to phase 3 of the GCM with Ketso methodology. The participants were recruited through snowball sampling (Peters, 2017). The 24 participants of phase 2 were invited and asked to recruit further participants from their own catchment partnership and flood risk management professional networks to widen the participation of practitioners to interrogate the findings of the study more widely. The aim of the workshop was twofold, first to share and consult on the mapped NFM delivery system according to the ways in which the practitioners approach and organise their knowledge production (Wietske et al., 2009) and second to interrogate the perception of the importance of individual barriers. Two separate working frameworks can be assigned to the two practitioner groups: flood risk management (Johnson and Priest, 2008) and integrated catchment management (Falkenmark, 2004). The statement lists, their ranking and the grouping of the barrier statements into concepts and their ranking were dis- cussed and reviewed by referring to these working frameworks as a catalyst to consider the socio-ecological-technical system as a driver for differences in perceptions and experiences.

Results
The 253 barriers generated in phase 1 were sorted and reduced to the final 47 (see Appendix A). The cluster map presented at the workshop consisted of seven concepts (Fig. 3b): politics, policy and planning, public perception, funding, infrastructure, evidence and technical knowledge. Group labels were selected in discussion with practitioners during the feedback and interpretation workshop aided by the labels generated through the sorting and ranking exercise.
The placement and appearance of concepts on the map indicate a degree of relatedness (Fig. 3a). Those concepts that are mapped in close proximity to each other are statements often grouped together by participants, revealing that they are conceptually similar, for example politics (Fig. 3b). Those mapped over a larger area or those that are more disperse signal that they are conceptually more heterogeneous, for example public perception (Fig. 3b).
An exploration of values and perceptions of the seven concepts that form the delivery system is made through ranking data, with a mean score calculated for each concept grouping (Fig. 4). The individual barriers that fall within each concept (see S1) were given a score according to the perception of importance. If ranked in the top 5 by participants, it received a score of 6, if placed between 5 and 10 a score of 5, between 11 and 20 a score of 4, between 21 and 30 a score of 3, and between 31 and 40 a score of 2, and the bottom 7 barriers received a score of 1. From this a mean barrier score for each concept is calculated (Fig. 4).
Both practitioner groups ranked the evidence and technical knowledge barrier concepts as being of lower importance as a barrier to delivery compared to the socio-organisational barriers of policy and regulation, politics and funding (Fig. 4).
The placement of statements on the map results from either participant's perception of similarity and indicates that participants have regularly sorted them into the same groups in phase 2 or by their difference, in that they are rarely grouped together in phase 2; in such cases multidimensional scaling places them on the map at a distance. However, when there are differences in understanding, perception, or interpretation of meaning by practitioner groups, the perceived difference can result in a statement regularly sorted by one group into a particular concept cluster whilst another group of actors recognises it as belonging elsewhere. In this situation these statements are described as bridging different concepts. The barrier statements highlighted with the red oval (Fig. 5) within the public perception and infrastructure concepts were thought to be a result of bridging different concept clusters: 33 (the long-term management and maintenance responsibilities are not clear), 43 (the timescales required for many NFM processes are long and do not match with our expectations of instant results), 11 (failure -who would be liable? The first failure will be disproportionately scrutinised), 2 (space -the interventions require a lot of space and the UK has a growing population), 16 (there is still a reluctance to use "new techniques" amongst the community. This includes the public, farmers, consultants and planners) and 44 (there is no general acceptance from funders that monitoring is needed). The participant interpretation and analysis workshop deemed that the bridging statements were an important concept and should be recognised within the concept map. The barriers were therefore grouped together as a new concept and renamed cross-cutting issues. This led to a redrawing of the concept maps (Fig. 6), including the loss of the concept infrastructure, with two of its barriers (16 and 44) moving to the new cross-cutting issues concept grouping. The remaining two infrastructure statements (24 and 28) were moved into the funding concept; they refer to flood defence payments and cost-benefit analysis respectively; therefore, a strong argument can be made for their placement within a funding group. Public perception became a smaller group of five barrier statements plotted closely on the map, showing conceptual similarity of the grouping. However, one  further barrier was moved from the funding concept, as the workshop participants thought it would be better placed in the public perception concept, supported by its close proximity to the public perception cluster (Fig. 5); this was barrier 34 (the public are not open to NFM techniques (leading to nimbyism): they resist change and feel safer with familiar options like dredging and have a poor understanding of risk and uncertainty). The changes were made as our methodological research approach is driven by knowledge co-production, including interpretation of results. The practitioners were of the opinion that these distinctions were useful for the wider transdisciplinary NFM study, and therefore barrier 34 was moved to the new public perception cluster. The Supplement data give the details of each individual barrier statement and the cluster which it was placed in following phase 3 of the GCM with the Ketso methodology.
The method revealed several individual barrier statements with large differences in perceived importance between practitioners contributing to catchment partnerships and flood risk and coastal management (FCRM) practitioners. Table 2 details the five with the largest differences.

Public perception
FCRM practitioners ranked individual barrier statements in the public perception grouping highly (S1), time (4), landowner power (5-6) and a resistance to change (7-10), the media (11-15) and how a landscape should look (11-15). Catchment partnerships have a distinctly different attitude to public perception as no individual public perception barrier ranked in the top 10; most received a middle ranking (see Appendix A). FCRM works with communities at risk of flooding; anecdotal evidence collated from flood-affected communities by Platt (2019) highlights a widespread public suspicion of authorities and a common belief that their particular flood event was contributed to, if not caused by, human error and the mismanagement of flood structures upstream. Flood risk management has been described as a blame game (Cowen and Delmotte, 2019;Krieger, 2013) where regulatory authorities must negotiate public scrutiny and accountability, further complicated by an ongoing debate as to whether responsibility for reducing flood risk lies with government, organisations managing flooding or society as a whole (Butler and Pidgeon, 2011;Klein et al., 2017). Additionally, a growing body of literature links social inequalities and living in flood risk zones (Fielding, 2018;Walker and Burningham, 2011), and it is against this backdrop that FCRM or any other organisation interested in expanding NFM to a mainstream flood management strategy must work; however, this context for delivery of NFM was not identified as a barrier by catchment partnerships, whilst it is not yet clear whether catchment partnership ranking of public perception as a barrier of lower importance is a result of their skill in public engagement or a lack of knowledge and exposure to the complexities of working with communi- Figure 6. The revised concept maps following phase-3 participant interpretation and analysis. ties living with flood risk or is driven by an alternative factor related to the conceptual heterogeneity of the grouping. The workshop discussion did reveal that catchment partnerships tended to see a reluctance to use techniques (statement 16) and long timescales for NFM processes (statement 43) as a barrier caused by unwilling or uninformed practitioners rather than the public. The GCM concept maps support this finding as catchment partnership practitioners placed these barrier statements within the technical knowledge concept compared with FCRM practitioners, who placed these barriers within public perception, a finding that suggests that FCRM practitioners do not perceive that they have agency to promote mainstream adoption; power lies with the public, who are not supportive. Public perception and the disparity in its perceived importance to mainstream NFM delivery are areas for further research.

Funding
The funding barrier cluster is ranked as having a greater importance as a barrier to NFM delivery by catchment partnerships than FCRM practitioners (Fig. 6); however, when individual barriers are examined, this broad distinction screens a more nuanced picture of funding and NFM delivery. For FCRM practitioners the difficulty in demonstrating a reduction of flood risk to households is ranked (5-6); for catchment partnerships this barrier is of much lower importance (33-34) ( Table 2). Catchment partnerships instead place staff resources (4), availability of funding (3) and difficulty in applying cost-benefit analysis (10-12) as barriers of much greater significance. These differences illustrate that the two practitioner groups are working within different funding systems from which to finance NFM delivery.
FCRM resources are principally derived from public funding with stringent conditions attached (DEFRA, 2011), one of which is to reduce flood risk to 300 000 homes by 2021. If the criterion to assign central government flood funding, reduction of flood risk to households, is not met by NFM, FCRM is unable to progress and support delivery as a mainstream flood management option. Host organisations of catchment partnerships are charitable bodies; the narrative is that this enables access to a wider number of funding sources. This flexibility makes them suitable for coordinating activities of a range of actors who access and contribute funds and resources to deliver activities of mutual benefit and according to local need (Bide and Cranston, 2014;Environment Agency, 2017). This study does not allow for an analysis of whether this is experienced in practice; however, the results suggest that catchment partnerships are experiencing difficulties in securing financial resources for NFM delivery.
Without a dedicated resource stream it is probable that NFM projects will be developed and delivered in an unsystematic ad hoc way strongly influenced by criteria set by the sources of competitive project funding. It is worth noting that catchment partnership practitioners do not rank a demonstration of a reduction of flood risk to households as important, indicating that there is a gap in their experience and/or understanding of conditions attached to central government flood funding.

Politics and policy and regulation
The most important barrier to the adoption of catchmentwide NFM according to both groups of practitioners is barrier 37, which describes a lack of commitment from government, particularly in conflicting and changeable messaging. The interpretation workshop (phase 3) gave an opportunity to explore this result further. To support this position, practitioners referred to a number of recent experiences to justify their perception that government did not actively support the adoption of catchment-wide NFM, including technical evidence ignored in the face of pressure from landowners in Somerset (Cowen and Delmotte, 2019), reversal on introducing statutory requirements for sustainable drainage (Wingfield et al., 2019) and that flood management is only a topic of concern immediately following a flood: "what we need is a good flood to get the attention of decision makers". However, practitioners also reported that the situation was in a state of change. Policy guidance encouraged the adoption of NFM, a single one-off NFM project fund of GBP 15 million had been recently allocated (Webb et al., 2018) and an NFM evidence directory (Burgess-Gamble et al., 2017) was due to be released. Practitioner feedback demonstrates a confused picture of government commitment to NFM adoption. Studies into the role of government in climate change adaptation have pointed to a failure of coordination between levels of government, from national to local, particularly with a lack of resources at the local level ; whether this applies to NFM delivery is an area for further analysis.
The policy and regulation concept displays the greatest divergence in opinions between the two practitioner groups. The most striking is barrier 13, "There are too many agencies, leading to poor communication, lack of coordination, inconsistencies and/or a lack of understanding". The catchment partnerships ranked this barrier at 8-9, whereas FCRM ranked it at 44 (Table 2). This represents a large difference in perception of importance. A similar idea to barrier 13 is captured in barrier 26, "It is difficult to work in partnership with so many organisations. There is not enough knowledge sharing, there is competition and there are differences in priorities", ranked 20 by FCRM practitioners and 8-9 by catchment partnerships. The difference in perception is not as great as barrier 13, but again greater importance is placed on this barrier by practitioners from catchment partnerships. From the GCM methodology it is not possible to say whether this is a result of an opinion of FCRM practitioners that they work well with other organisations or whether the ranking is driven by the opposite position that cross-sectoral working is not important. Given that the Environment Agency's own internal review (Table 1) conducted in 2019 concluded that a culture change within flood risk management is required to open up collaborative working, this suggests that perhaps the latter is most probable. Instead of approaching NFM delivery from a catchment-based perspective, FCRM practitioners are attempting to fit NFM delivery within the traditional engineering-led approach, are unaware of or resistant to the integrated nature of NFM delivery, and do not agree that other agencies are needed to adopt NFM as part of a mainstream flood management strategy. Catchment partnerships on the other hand were established to bring together interests of different land and water management bodies (Cascade, 2013); as such they would be expected to have a good under-standing of whether different sectors and organisations are well aligned to facilitate delivery and greater collaborative working required to support NFM delivery. The high ranking of barriers 13 and 26 suggests they value collaborative working but that this is not operating effectively. Whether this is unique or particular to the NFM delivery system or other activities and governance of catchment partnerships more broadly is an area for further research.

Evidence and technical knowledge
Both flood risk authorities and catchment partnerships recognise that because of the large number of physical processes involved in NFM, there are scientific uncertainties in their optimal use. However, barriers within the technical knowledge concept notably have not received a high ranking by either practitioner group (within the top 20 -S1). Conversely, during the interpretation, workshop practitioners, in particular those from FCRM, did not agree that the evidence and technical knowledge concept group did not act as an important barrier to delivery. There was an insistence that it should be understood as one of the most important barriers. To support this perception, practitioners described a technocratic linear process of first filling evidence gaps, leading to policy change which then opens up funding streams. This position was somewhat contradicted when the workshop went on to discuss the number of supportive policy documents that have been in circulation for over a decade. This discrepancy of whether there was sufficient evidence to formulate policy or whether there was a failure in policy delivery could be explained in part by the practitioners who engaged with the project. They are technical experts who are responsible for delivering land and water management interventions, conceptualising that task through a natural sciences outlook. If their rating of the importance of individual barriers was influenced by familiarity, their expertise and comfort with accessing and using the technical evidence base are likely to lead to a lower ranking of those barriers compared to social or political barriers. The participants identified this in the interpretation workshop and called for greater inclusion of senior managers responsible for strategic planning and policy makers in any further research.

Cross-cutting issues
The group of cross-cutting issues contains barriers described by one participant at the interpretation workshop as the "parts that don't belong to anyone and get easily forgotten". These barriers have been described as bridging statements, indicating linkages or interconnections between concepts. When phase-2 data were reviewed to see which concepts the statements were bridging, it was only possible to identify this pattern for two barrier statements: 16 (a reluctance to use new techniques) in which participants had grouped the barrier most frequently with both public perception barrier state-ments (34, 35 and 36) and a technical knowledge statement (5). Barrier 43 (long timescales for NFM processes) had been grouped into both technical knowledge (12 and 19) and public perception (25 and 34). Within the cross-cutting group these are also only barriers that were ranked highly (see Appendix A). The remaining barrier statements in the crosscutting group, 2, 11, 33 and 44, did not show any strong association with any of the other barriers. Why these particular barriers were not linked to other barriers within the identified 47 is unclear. It could be that their meaning was considered ambiguous or represents components of the delivery system that this group of practitioners was unfamiliar with, which would also explain their low importance ranking.

Lessons from applying the approach in practice
The GCM with the Ketso method worked well: outputs combined with the interpretation workshop have allowed an exploration examining interconnections and perspectives of practitioner groups, identifying problems for further investigation. The group concept mapping method allowed for a relatively large number of practitioners to input for this type of knowledge co-production study -89 individuals across three phases. However, the portion of the study that produced the basic concept maps was limited to the opinions of 24 individuals. The conceptual framework dividing the NFM delivery system into seven areas should be tested in practice and reviewed against published theories. A number of findings, particularly that the greatest constraints on delivery come from socio-organisational factors, support the conclusions of studies examining barriers to the implementation of linked environmental management policies, such as climate change adaptation Moser and Ekstrom, 2010) and blue-green infrastructure (O'Donnell et al., 2017;Thorne et al., 2018), and it is thought likely that the seven concepts underpinning NFM delivery are translatable nationally and internationally.
The interpretation workshop is a key component of the group concept mapping methodology; however, it must be understood as a snapshot in time in what is an evolving area of research and practice. A number of changes to the concept maps recommended in the workshop were implemented: while the alterations did not change outputs for individual barrier statements, they did result in changes to the mean rank importance score of public perception, identifying it as the greatest barrier concept grouping to NFM delivery for FCRM practitioners. As discussed above, the newly formed group called cross-cutting issues revealed uncertainty as to whether the bridging nature of the statements is driven by practitioners' lack of familiarity or by ambiguity in phrasing.
Research on risk and uncertainty shows that there is a difference in perception within organisations from practitioners with differing authorities and responsibilities (Höllermann and Evers, 2017). Inclusion of a greater number of senior managers with experience in strategic planning and policy making would have enabled a comparison of the perspectives from this actor group. In the wider transdisciplinary study there may be a need to revisit the problem identification phase with this group of actors (Pohl et al., 2021).

Conclusion
The perceptions of barriers to NFM delivery between two actor groups, FCRM and catchment partnerships, have been examined and compared as a problem identification phase of a wider transdisciplinary study. Transdisciplinary studies begin with problem-framing and structuring perspectives of practitioners and researchers across disciplines and sectors to develop mutual understanding as a basis for further research and cooperation. Topics for further analysis and key findings of the problem-framing phase are public attitudes to NFM techniques, cross-sector collaboration and partnership working, FCRM practitioner culture and the question of what a successful outcome is (clarity from central and local government messaging).
There is broad agreement that the UK government is not sufficiently supportive and is impeding delivery of NFM. Whether this is a problem of central government and national policy making, is associated with local government or is found across all administration levels of government remains unclear. For FCRM practitioners, public perception was the most important barrier to NFM, and for catchment partnerships it was policy and regulation. The findings of the group concept mapping methodology support the literature on barriers to environmental management policy adoption more widely, that the greatest constraints on change come from socio-organisational factors rather than a lack of technical knowledge; however, it must be acknowledged that practitioners queried this finding in the interpretation and analysis workshop. The implications of these findings are significant, as the areas identified within this study as being of greatest concern are those that often receive the least research funding. Current research efforts often focus on approaches and attempts to improve technical knowledge; this in part may reflect the nature of research funding as operating within disciplinary boundaries, but for NFM to be successful further system-based research is needed to move beyond disciplinary boundaries and reflect the integrated nature of NFM delivery.
The group concept mapping methodology has been employed as a problem identification method allowing for the active participation of a multidisciplinary group of 89 practitioners, from which the system of NFM delivery has been revealed. This holistic process has allowed for an interrogation of different perspectives between practitioner groups who have different working conceptualisations and knowledge systems. Two separate working frameworks can be assigned to the two practitioner groups: flood risk management (Johnson and Priest, 2008) and integrated catchment management (Falkenmark, 2004). Beyond the context of UK NFM delivery, the approach described in this paper can be utilised across the globe to support the delivery of NBS for flood management in providing a framework for systematic problem-oriented research. NBS as a practice-based approach that explicitly includes a second pillar of social safeguarding within traditional conservation efforts (Cohen-Shacham et al., 2019;Ruangpan et al., 2020) requires an expansion in knowledge about how to implement them across scales, contexts and people. Concept maps create visual representations of the networks of ideas to describe how a group thinks about an issue (Goldman and McLinden, 2013), allowing for problem-oriented integration of different knowledge bases across sectors and disciplinary framings. The maps expand the resulting knowledge beyond the context in which it was generated by encouraging concepts to be deconstructed, to be restructured or to emerge (Rosas, 2017).
An interpretation of the study is that the FCRM sector is resistant to change; a means of lessening isolation from other sectors can be achieved through social learning via knowledge co-production. In this respect the study approach has been particularly impactful: participating practitioners developed mutual understanding, constructed a map of concepts underpinning the NFM delivery system and explored differences in perceptions impeding delivery. This study is the first step towards identifying means of encouraging the adoption of NFM as a mainstream flood management technique using a transdisciplinary approach of problem framing with NFM delivery practitioners.
Appendix A Table A1. Forty-seven individual barrier statements generated via phase 1. Qualitative statement generation, sorted into seven concepts in phase 2, mean importance and rank according to practitioner groups. The top 10 barriers are underlined to highlight them. Those in bold typeface indicate a large difference in perceptions between the two actor groups. There are too many agencies, leading to poor communication, lack of coordination, inconsistencies 3.3 44 5.5 8-9 and/or a lack of understanding. 14 Poor integration of planning and design policy on a catchment scale, strategically and politically 5.5 7-10 5.8 5 17 Locked in -standardised approaches and procedures mean regulators are locked in and are inflexible; 4.3 28-29 5.6 6-7 for example, it is difficult to deliver projects with multiple benefits. 21 Current organisation (or lack thereof) means we are in danger of a scatter-gun approach. 5.3 16-17 4.9 27-28 26 It is difficult to work in partnership with so many organisations. There is not enough knowledge 5.15 20 5.5 8-9 sharing, there is competition and there are differences in priorities.

38
Some regulatory procedures make adoption difficult, e.g. the reservoirs act, countryside stewardship agreements, and the common agricultural policy Organisations do not resource staff to do these types of projects; they can be on short-term 4.5 27 5.9 4 contracts, so there is no consistency, or they do not have the capacity or time. 24 Flood defence payments are based on reducing flood risk to households. It is difficult to 5.6 5-6 4.7 33-34 demonstrate NFM contribution to this. 28 Option selection is currently based on cost-benefit analysis. The benefits and costs of NFM are not 4.9 23-25 5.4 10-12 understood, and so it is excluded. 40 Funding -insufficient, difficult to access and inappropriate; for example, will not pay for staff time 5.3 16-17 6 3 or does not join other funding streams. 42 The efficiency culture is to reduce spending, not to maximise outcomes, and NFM is not always the lowest cost option.  Data availability. Appendix A gives details of the qualitative statements generated through the participatory workshops. These statements are the underlying research data that were ranked and analysed in the subsequent steps.
Author contributions. TW was the lead author and creator of the images and the researcher of the study responsible for conceptualisation, design of the methodology and formal analysis. She also managed the timing and logistics of the investigation and data curation. NM supported TW in theoretical conceptualisation, data validation and writing -review and editing. He provided doctoral supervision and funding acquisition. KP contributed to the study and production of this paper in supporting data validation and authoring a number of sections and editing and review of the complete paper; she provided doctoral supervision. JS was instrumental in funding acquisition and recruitment of participants. He reviewed and edited earlier drafts of the paper and was a doctoral supervisor.
Competing interests. The contact author has declared that neither they nor their co-authors have any competing interests.
Disclaimer. Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Special issue statement. This article is part of the special issue "Contributions of transdisciplinary approaches to hydrology and water resources management". It is not associated with a conference.