Who Is Connected with Whom? A Social Network Analysis of Institutional Interactions in the European CCA and DRR Landscape

: Communication and collaboration are critical for designing and implementing responses to climate change impacts and related disasters. This acknowledgement has increased interest in understanding social and institutional networks for climate change adaptation (CCA) and disaster risk reduction (DRR). In this study, we used Social Network Analysis (SNA) to explore institutional interactions within and across the communities of the aforementioned domains in Europe. Firstly, we investigated the type and intensity of interactions. We calculated SNA metrics to assess the roles of di ﬀ erent actors and applied cluster analysis to identify actors with similar patterns of connections. SNA showed that communication is often more intensive within the two communities, while collaboration is frequent around topics related to both CCA and DRR. Cluster analysis revealed that actors tied with DRR were more closely connected, while actors tied with CCA and those with mixed connections showed no obvious clustering a ﬃ nity. The European Climate Adaptation Platform, Climate-ADAPT, had the highest value for various SNA metrics, reﬂecting its popularity in the network and its potential for enhancing interactions among its actors. Finally, SNA was complemented by qualitative interviews, which emphasised the importance of connecting CCA and DRR in organisational mission and vision statements.


Introduction
After almost two decades of ongoing discussions and negotiations, adaptation has become a priority in national and international political agendas, and is now considered not only an important but also a necessary element of any autonomous or planned response to climate change [1][2][3]. Similarly, past years witnessed growing efforts to improve reduction of disaster risks, by looking at their causal factors and possible future risks, paving the way for an integrated framework for disaster risk reduction [4].
Climate change adaptation (CCA) and disaster risk reduction (DRR) both aim to address issues relevant for multiple sectors and governance levels, such as the changing type, severity, and frequency of climate change impacts, and weather-and climate-related risks. Such challenges require flexible and indicate the roles that different actors have in a network, to identify which actors are better connected and through which type of interaction [35], and to visualise the emerged networks via sociograms.
Qualitative information collected via interviews with key actors, as done in this study (see Section 2.2), may complement the findings of an SNA by providing contextual information [36] that helps to better understand why certain interaction patterns emerge. Such double-step approaches are clearly desirable for results' triangulation. Nevertheless, they are rather uncommon due to the additional time and effort required [37].

Data Collection
An inventory of actors active in the fields of CCA and/or DRR was prepared prior to the application of the SNA. The inventory included information on the (1) vision/mission statement, (2) core competencies, (3) field of work (CCA, DRR, other), (4) sector, (5) geographical scope of work, (6) target group, (7) physical headquarters, (8) URL address, (9) contact person, and (10) other information. As of April 2016, 137 stakeholders (stakeholder and boundary organisations, knowledge platforms, policy and research initiatives, existing networks/partnerships, and end user needs (SPINE)) had been identified through extensive desktop research, expert judgement and snowball sampling, providing a good overview of actors from the CCA and DRR communities visible on the European level. The two latter techniques were used to complement the initial list of key actors obtained through desktop research. During several CCA and DRR events, a set of relevant actors was asked to indicate other actors with whom they have a professional relationship with regard to their work on CCA and/or DRR issues. The procedure was repeated with each group of the identified stakeholders until the research team was sufficiently confident that all major CCA and DRR stakeholders active in the European arena were covered. Methods such as expert judgment and snowball sampling may introduce a certain degree of bias due to their inherent subjectivity. In this study, however, saturation was reached; a point where no new actors were suggested to be contacted. This indicated that all key actors had been identified and considered in the study.
Subject to our objective, 35 actors were selected from the above-mentioned list using expert elicitation method, to identify a compact but representative sample of the European CCA/DRR landscape. All actors included in the final set were invited to participate in the SNA survey (Appendix A).
SNA was conducted using information about the intensity and the type of interactions between the selected actors (the survey questionnaire can be found in Appendix B). Data were collected through an online survey that was developed on the SurveyMonkey platform and administered by email to one contact person for each actor considered in the network. Potential respondents were selected based on their suitability for answering the specific questions. In some cases, invited respondents suggested alternative people to participate in the survey on behalf of their organisation. Whatever the case, participation in the survey was by invitation only and responses were collected from only one representative of each of the selected actors. Invited participants were encouraged to consult with their colleagues, especially in cases where the organisations they represented had more than one unit working on CCA-and DRR-related issues. Data collection took place in July and August 2016, and 32 responses were received out of the 35 invited actors. For confidentiality reasons, this paper does not contain any information on the actual respondents except for the names of the organisations. The electronic files of the responses can be made available upon request and only after asking for the consent of the organisations involved in this exercise.
After the SNA exercise, in-depth interviews were conducted with a subset of the top-performing actors according to SNA metrics (see Table 2 in the 'Results and Discussion' section and Appendix C). These actors were informed about the purpose of the follow-up interviews and received a confidentiality statement along with a compilation of the SNA results prior to the interviews.
The interviews aimed to explore whether the SNA results were in line with the expectations of the respondents or not, and also collected additional information that could support a better explanation of the SNA results. Follow-up questions explored additional aspects, such as the factors that positively influence openness towards the other community and the perception of their own and the counterpart community. Given the above-mentioned aims, interviews followed a semi-structured approach, using a predefined set of closed and open-ended questions (See Appendix D for the interview guidelines). This approach allowed us to maintain the focus of the interview, providing at the same time sufficient space for additional questions, comments and reflections.

Data Analysis
NodeXL Pro software package (Version 1.0.1.389) (http://www.smrfoundation.org/nodexl/) was used to calculate SNA metrics that show the basic properties of the network under study (Table 1) and generate sociograms to visualise the network's communication and collaboration interactions between different actors.
In the SNA, we considered all interactions except for those indicating awareness of actors but lack of interaction with them (level 1 -"I am aware that this actor exists, but to my knowledge, we do not have any contact with them", see Appendix A). Directed sociograms were generated using the Haren-Korel Fast Multiscale layout algorithm, "a force-directed layout algorithm using optimisation in order to make the algorithm computationally efficient" ( [38], p. 274]).
Besides the SNA, we also conducted a cluster analysis to detect groups of actors with similar patterns of connections, by minimising the diversity within a cluster and maximising the diversity between them. The NodeXL group analysis tool provides the Clauset-Newman-Moore algorithm and the Girvan-Newman algorithm. The Clauset-Newman-Moore algorithm is an agglomerative algorithm; it starts with a community partition, where each single vertex represents a community and then at each iteration, a pair of communities is merged into a single one, improving the measure of cluster quality. This process continues until any further aggregations would decrease the overall quality [39]. The Girvan-Newman algorithm is a divisive method based on the edge removal; it progressively removes edges from the network, focusing on edges between loosely connected communities (i.e., those with high "betweenness" value). By removing these edges, the communities are separated from one another and presented as clusters [40]. Table 1. Definitions of the SNA metrics calculated in this study.

SNA Metric Definition
Density It is calculated as the proportion of the total number of actual interactions among the actors of a network, out of the total number of potential interactions. The higher the value of the density, the higher the connectivity of the network.

Degree centrality
It presents the total number of interactions of an actor regardless of their direction. It is a good indicator of certain actors' ability to communicate directly with other actors in the network, as well as of their importance for transferring information. In the case of directed networks, degree centrality is divided into in-degree and out-degree.

In-degree centrality
It presents the total number of incoming interactions to an actor. Actors' in-degree centrality increases with the number of actors indicating that they seek information or collaboration with the former and does not consider an actor's own assessment. It may be used for identifying actors that are popular in a network.

Out-degree centrality
It presents the interactions outgoing from an actor. It may be used for identifying actors that tend to interact a lot with other actors in the network as a result of the initiative of the former.

Betweenness centrality
It is calculated on the basis of the times that a vertex acts as a bridge in a network (i.e., lying on the shortest path between two actors). An actor with a high betweenness centrality value has the power to connect disconnected groups, to broker opinions and to influence the flow of information. It may be used as an index of the control that an actor has over the connections in a network.

SNA Metric Definition
Eigenvector centrality It is calculated on the basis of an actor's own degree along with the degrees of the actors to which the former connects. It assigns relative scores to all vertices in a network based on the principle that connections to a high-scoring vertex contribute more to the score of the vertex in question than an equal number of connections to low-scoring vertices. It is another measure of an actor's importance in a network and may be used for identifying the popularity of an actor's partners.

Clustering coefficient
It quantifies the proximity of a vertex and its neighbours. It is calculated on the basis of the degree to which vertices tend to cluster together. Vertices with high degree values have lower clustering coefficients because they connect to many vertices who are not connected among them. On the contrary, vertices with high clustering coefficients tend to have fewer connections, since small groups have larger potential to connect every member to one another.
Finally, a content analysis of the interview notes, including coding and qualitative assessment of the collected information, was carried out to identify concepts/themes contained in the data. Excel filters helped to analyse the data vertically-looking for coherence and a narrative thread within each interview, and horizontally-trying to recognise general patterns and differences.

Social Network Analysis
Social network analysis considered 35 vertices (i.e., actors) and a total of 470 edges (i.e., interactions). For the three actors that did not participate in the survey, we considered the relationships that other actors in the network specified to have with them. Among the survey participants, 16 actors had CCA as their primary or secondary core competency, 11 came from the field of DRR (including risk assessment), while the rest of the actors came from the insurance, climate research, and sustainable development sectors, with direct links to the field of CCA and/or DRR. All of the actors represented leading organisations that operate in, or have an impact on, the European CCA and DRR landscape, and have different but overlapping mandates, roles, responsibilities, and interests (information on actors' characteristics can be found in Appendix E). Table 2 provides an overview of the top performing actors based on the calculated SNA metrics. Although a few "outliers" could be observed scoring high in only one category, altogether, the highest places in the ranking with regard to different SNA metrics were very often occupied by the same actors. These included EU institutions in charge of EU policies (e.g., DG CLIMA, DG RTD), EU organisations involved in knowledge production and dissemination (e.g., JRC, EEA), international bodies (e.g., IPCC, UNEP, UNISDR-Europe), research institutes with international reputation (e.g., IIASA), networks (e.g., EpaNet, STAG), and platforms (e.g., Climate-ADAPT, PreventionWeb). The calculated scores reflect the popularity of these actors and their high potential for improving information flow and links, especially with actors that are not well connected to the rest of the network. High values were calculated also for actors coming from the private sector such as insurance companies (e.g., MunichRe, Swiss Re) for all but one metric (betweenness centrality). This result reflects the transition that is seen in many countries where the influence of the private sector overall both on the development and implementation of CCA and DRR policies has been growing constantly and with prospects to increase further in the future (https://climate-adapt.eea.europa.eu/eu-adaptation-policy/strategy). Still, the lower betweenness centrality of these actors may signal that they are not yet ready to play a significant role in connecting other actors in the merged network (or are not yet recognised as such).
An interesting observation is that many actors had high eigenvector centrality values. This likely resulted from the fact that they are connected to the 'well-connected' actors, such as Climate-ADAPT and DG CLIMA, increasing their overall score for this metric. Furthermore, this finding emphasises the role of the best-connected actors in reaching out to and bridging other actors in the network. The European Climate Adaptation Platform (Climate-ADAPT) emerged as the actor with the highest value for most SNA metrics: i.e., degree (Figure 1a), betweenness (Figure 1b), and eigenvector centrality ( Figure 1c). These findings reflect the popularity of Climate-ADAPT in the network, its important role in communicating with the rest of the actors (i.e., degree and eigenvector centrality), as well as its outstanding ability to set up communication bridges with those actors that do not have reciprocal interactions (i.e., betweenness centrality). With regard to the in-degree centrality, which indicates the number of ties others specified to have with a specific actor, Climate-ADAPT was positioned only after DG CLIMA, EEA, IPCC, and JRC. Further to Climate-ADAPT, similar disparity between in-and out-degree centrality values was observed also for DG RTD (22 and 18 for in-degree, against 28 and 29 for out-degree, for Climate-ADAPT and DG RTD, respectively). This higher number of the outgoing compared to the incoming interactions, may imply that Climate-ADAPT and DG RTD seek, in an active way, to reach out to other actors in the network, which is seen as a way to engage with their target audiences. Such an attitude and the position in the network, as indicated by the high betweenness and out-degree centrality, point out these two actors as good candidates for linking other actors in the merged CCA and DRR network. The important and influential role of Climate-ADAPT that was revealed from the SNA in this study agrees with findings from other recent studies that have highlighted the potential of web platforms to support an improved communication and knowledge exchange between the CCA and DRR policy domains [2,19,44]. Climate-ADAPT was launched in 2012 by the European Commission and the European Environment Agency (EEA) in an attempt to facilitate dissemination and use of the produced 'adaptation knowledge' in Europe, and through this to support evidence-based policy-and decision-making, and improve coordination among actors operating in different sectors and at different governance levels [45]. Although the primary focus of Climate-ADAPT is indeed the field of CCA, it is worth noting that the amount of DRR information that it provides has been constantly increasing. For example, as at 22 January 2020, the Climate-ADAPT database included 553 items relevant to the DRR sector, while a page providing information about the policies of the DRR sector is available on the main site of the platform. Moreover, the recent evaluation of the EU Adaptation Strategy [46] also included the evaluation of the Climate-ADAPT platform and confirmed its increasing value for several types of users (e.g., policy makers, city planners, experts). The outcome of the SNA in this study seems to be confirming this conclusion.
Similarly, the Directorate-General for Research and Innovation (DG RTD) has a mission to develop and implement research and innovation policies in Europe. Tackling some of the main current and future societal challenges has been identified as one of its key aims. Clearly climate change, climate change impacts, and weather-and climate-related disasters are among the most important and direct challenges that societies are facing. Therefore, this may explain the fact that DG RTD has an active role in reaching out to both of CCA and DRR communities.
Further to the calculation of the SNA metrics, a graphical representation of the SNA results was prepared in the form of sociograms. As actors in the investigated network were densely connected (network density: 0.39), the following sociograms illustrate only the strong interactions, while the more sporadic ones were excluded in order to improve visualisation. More specifically, Figure 1a-c depict the network of actors that have established (either by indicating or receiving) frequent communication and frequent or institutionalised collaboration interactions with other actors in the network. The different colours of the ties reflect the type of the interactions, while node size reflects the values of the SNA metrics; degree, betweenness, and eigenvector centrality, calculated for each actor.
Finally, taking a closer look at the type of interactions that actors indicated (Table 3), we saw that the type of interaction was specified by participants in almost all cases (ca. 90% for communication and 95% for collaboration interactions, respectively). Collaboration was most often mentioned with reference to issues that relate to both CCA and DRR, while communication was more frequently mentioned for issues that refer to the CCA domain. This result might suggest that issues related to either CCA or DRR can be tackled effectively through communication channels alone, while addressing issues that are relevant to both domains requires closer interactions among interested organisations. As already discussed in the literature, insufficient or lack of cooperation across responsible institutions is a key constraint when facing a crisis or emergency in different contexts [47]. For this reason, the presence of collaborative institutional interactions is usually considered a positive element in a network. However, there are conditions where collaboration may not ensure successful responses [48]. A holistic assessment of collaborative interactions is expected to be more insightful in showing their usefulness and impact on the overall network and the processes taking place in it (e.g., learning) [47][48][49]. This assessment should not only consider numerical information (i.e., total number of indicated interactions), but also other dimensions, such as the quality and efficiency of collaboration.
Finally, cross-organisational communication and collaboration are indeed crucial for addressing issues relevant to both CCA and DRR policy domains. Although this study provides evidence that suggests that such collaborative interactions have been already taking place, a full integration of CCA and DRR is still to be achieved in the future.

Cluster Analysis
Cluster analysis, using the Clauset-Newman-Moore algorithm, resulted in two large and one small cluster (Figure 2a). This visual representation of the cluster analysis shows that connections related to DRR predominantly appear in one of the two large clusters. Then again, connections related to CCA appear mainly in the other large cluster and in the small cluster, but they are also present in the "DRR dominant" large cluster. Cluster analysis was repeated using the Girvan-Newman algorithm based on edge removal. This clustering exercise found one large and 21 small (one-member) clusters ( Figure 2b). The large cluster contains mainly DRR actors, while no obvious clustering pattern was found between the actors related to CCA or those with mixed connections. Overall, the cluster analysis with either of the two different methods shows that actors tied with DRR are more closely connected, which may suggest the existence of more effective interactions among them. Addressing matters related to DRR is characterised by a certain degree of urgency, which requires fast responses and, therefore, closer and more structured interactions. On the other hand, the fact that actors tied with CCA and those with mixed connections showed no obvious clustering affinity may reflect the fact these types of connections involve a more widespread network of actors, as they deal with many different topics, sectors, and policy levels.

Key Results from the In-Depth Interviews
In-depth interviews were conducted with representatives from Climate-ADAPT, DG CLIMA, DG RTD, EEA, Intergovernmental Panel on Climate Change (IPCC), Joint Research Centre (JRC), PreventionWeb, and UNISDR Europe (one interviewee represented two organisations). Organisations were selected based on the importance of their role in the investigated network as this was reflected in the calculated SNA metrics (see Table 2).
Interviewees' expectations regarding the roles and the interactions of the considered actors were in agreement with the SNA results, but for a few cases. For the case of Climate-ADAPT, the popularity of the platform (i.e., high degree centrality value) was explained by the fact that it provides a wide range of information, reports, tools, and case studies, making it an important source of information for stakeholders operating at different governance levels and in different sectors. Its important role as a bridge between CCA and DRR communities was related to its mandate and the actions taken by its governing bodies to support and contribute to the coherence between the two domains. As a result, relevant initiatives taken by the platform itself, e.g., inclusion of DRR material in the database of the platform, as well as other relevant actions taken by its governing bodies towards the achievement of this objective (e.g., the EEA published in 2017 a report on CCA and DRR in Europe [18]) were mentioned to justify the role attributed to the platform. Also, some interviewees stressed that Climate-ADAPT actively reaches out to interested stakeholders by means of diverse activities and channels (e.g., workshops, conferences, newsletter with over 5000 recipients), which agrees with the high out-degree centrality score that was calculated for this actor.
When asked about the organisational characteristics and other external factors that can positively influence an actor's openness towards the other community, interviewees most frequently named the role of the organisational strategy that creates a sense for collaboration and previous experience with international and interdisciplinary issues. Commitment and determination to conduct joint projects, as well as strong leadership and support from the top management, were ranked second with regard to the frequency of responses. Tangible resources such as financial and human resources and IT were also mentioned as important enablers for creating open interdisciplinary work environment. Nevertheless, such factors were referred to less frequently, underlining the role of "soft" features (see Table 4 for the full list of factors). Table 4. Factors enhancing organisational openness and interdisciplinary work according to interviewed stakeholders (numbers in the parentheses indicate the number of respondents who mentioned each factor during their interviews). With regard to the results of the cluster analysis, respondents confirmed unanimously the existence of two clusters related to CCA and DRR, respectively. Interpretation of the respondents' results could be traced back, to some extent, to the perception question, in particular with regard to self-perception and perception of the other community. Respondents working in the field of CCA perceive DRR as a single solid cluster with a long tradition, having its roots in humanitarian aid, and describe CCA as a relatively younger and fast-developing domain, that deals with many topics relevant for several sectors and disciplines. On the other hand, interviewees with a DRR background describe CCA as a part of the DRR policy domain, since climate change is just one out of many risks that DRR deals with. One interviewee referred to mitigation and adaptation as "[ . . . ] a form of risk reduction", since it aims to reduce the occurrence of new and limits the amplification of existing climate-related risks. Furthermore, actors with expertise in DRR perceive their community as an established field with a great deal of practical knowledge in the area of emergency response, spanning over social, health, and earth sciences. However, the production of science-based knowledge is considered a relatively new task in the DRR community, triggered by increased interest in planning and prevention.

Factors Enhancing Organisational Openness and Interdisciplinary Work No of Responses
In contrast to CCA respondents, the interviewed DRR experts were surprised by the difference in the density of the two clusters. One interviewee pointed out that in reality the DRR community is more widespread as it includes many small, less-institutionalised knowledge producers, further to the organisations that are embedded in the field in strict terms. The same interviewee pointed to the fact that by selecting the most prominent DRR actors from the European/international level, the SNA managed to capture only "a piece of the DRR universe". Another interviewee coming from the DRR community hypothesised that the spread of the CCA community might be attributable to its need to draw on a wider knowledge spectrum, than the one used by the DRR community, which depends heavily on future projections and requires a lot of simulation work and modelling.
Furthermore, all respondents pointed to a lack of alignment between CCA and DRR. This originates to overcome these challenges will depend very much on the context and specific design of any other study. Nevertheless, communicating the key constraints faced here may help other researchers when designing analogous SNA exercises in the future.
(a) Boundaries of a network: Deciding on the composition and size of a network is a common challenge in SNA. An ex ante attempt to create a balanced network in this study proved to be rather difficult, as certain actors could fall in more than one category of the classifications that we considered. Also, looking at the CCA and DRR communities considered in this study, some of the interviewed experts commented that both of them (especially the DRR community) are more diverse in real life. Involving a larger number of actors may have ensured a wider and perhaps more accurate representation in the network. Nevertheless, such an approach could have other implications, such as a negative impact on the overall response rate of the survey (see point c below). (b) Survey administration: Identification of suitable survey participants, frequent staff turnover, and unavailability of some invited participants to provide responses on behalf of their organisations were the main reasons that delayed the prompt administration of the survey. (c) Repetitive pattern of the SNA questionnaire: Questions in the SNA survey had a rather repetitive pattern. Especially in cases of large networks, this characteristic causes fatigue, which is usually responsible for incomplete or lack of responses. In our study, participants were asked to specify both the type and intensity of relationships they had with other actors, causing an additional workload while responding to the survey. (d) No responses: Interactions of the actors that did not participate in the survey were treated as 'no responses', allowing us to take into consideration the interactions indicated by the survey respondents also for the non-respondents. Although it was not the case in this study, this approach may have implications on the calculated SNA metrics (see also [29,33]), especially if the non-respondent actors receive a large number of interactions. (e) Subjectivity bias: Survey participants were encouraged to consult their colleagues when answering the SNA survey, especially when more than one division was involved in CCA and/or DRR issues. It was not possible, however, to control if answers were achieved unanimously. This might have introduced a self-perception bias in the SNA, since respondents' subjective responses provide the factual input to the interaction analysis. Using alternative methods, such as focus groups, could potentially limit this barrier. (f) Interpretation of SNA results: SNA is a suitable tool for exploring social structures and the interaction patterns among different actors. Nevertheless, it has limited potential when trying to explain the reasons why such structures or patterns exist, why actors have certain roles, or relate to each other the way they do. Complementary use of the SNA with other methodological tools (e.g., in-depth interviews as seen in this study) can result in a more insightful interpretation of the results that emerge from its application. Similarly, the design of multi-level SNA exercises, which combine 'whole-network' and 'ego-network' analyses (see [29,30]) has the potential to provide a better understanding of the networks both in terms of the roles of the actors they entail as well as of their interactions. (g) Network dynamics: SNA exercises, such as the one presented here, are rather static and can capture only a snapshot in time. At the same time, networks (i.e., actor interactions and the roles they have in a network) are rather dynamic and responsive to changes in their composition and the overall context in which they operate. A regular repetition of such exercises [29,36], both the 'monitoring' of the preliminary inventory of actors and the SNA exercise itself, can help understand how networks evolve over time.
In spite of the above-mentioned challenges, the value and usefulness of the SNA should not be undermined. SNA has the potential to provide objective (quantitative) information on how actors are interrelated and the roles they play in networks through systematic and scientifically robust methods. Such information can be used to inform a practical strategy to support the improvement of the interactions and the coherence between different communities of interest.

Recommendations
Even if efforts to close the gap between the two communities have already been implemented, evidence of an insufficient alignment of the two fields is still present. Existing barriers restrict a more effective communication and collaboration between the two communities, which are seen to operate to a large extent in silos. SNA results along with the information collected through the in-depth interviews allowed for the formulation of a tentative set of recommendations for improving actors' collaboration, subject to further empirical testing: (a) Coherence between the two communities can be achieved through knowledge sharing.
Engaging in this process actors with high out-degree centrality values (e.g., Climate-ADAPT and DG RTD), which indicate their strong interest in outreach activities, could have a positive impact on knowledge dissemination, contributing to the development of stronger ties between the two communities. (b) The exercise presented in this paper was meant to foster dialogue and support partnerships among stakeholders with similar or complementary interests. Feedback received during the survey confirmed the achievement of this goal and highlighted the need for closer cooperation. (c) The importance of having diverse opinions in CCA and DRR dialogues has already been acknowledged, especially when thinking of ways to enhance synergies between different communities at stake [1]. Although it has been observed that actors seek to interact with other actors with whom they share some similarities (i.e., homophily), ensuring diversity in a network is expected to be of benefit in terms of its potential to expand, increasing its strength and relevance in the longer term. (d) Future research should focus on actors that have key roles in their networks (e.g., actors with high potential for setting direct interactions with other actors in their networks, and thus for sharing information, or others that have the ability to connect actors that are not well connected with the rest of the network) in order to better understand which of their characteristics or competences have enabled or facilitated their interactions with other actors. Similarly, it is important to identify and investigate the profiles of actors that have a low number of connections, in an attempt to develop solutions to overcome this barrier. Egocentric SNA, as seen in other studies [30], might be a useful approach for this purpose, as it provides useful insights in the type of interactions of distinct actors. (e) Involving actors that are often neglected in relevant exercises is important for multiple reasons.
Perhaps the most important reason is that placing less known, yet important, actors on the map of the CCA/DRR landscape increases their visibility. This is the starting point for any type of interaction; either by providing a simple awareness of their existence by other actors, setting the ground for the beginning of their communication with others, or in a more optimistic scenario, helping them establish strong and fruitful collaborations. It is acknowledged, however, that introducing new actors in established networks is not always a trivial task [50] and may require systematic efforts before such changes are achieved and fully accepted. (f) Finally, efforts should be placed on emphasising and promoting the complementarity of both different fields and resulting benefits through a joint resilience narrative. This would support the development of a mutual understanding and foster shared goals and networks of collaborative actions. In the case of the CCA and DRR communities, institutions that aim to fill in existing gaps, overcome barriers, and, finally, bring the two communities closer need to have strong motivation and commitment, and to show patience and perseverance, as the alignment of interests is a rather time-consuming process with a considerable risk of failure.

Conclusions
Climate change adaptation (CCA) and disaster risk reduction (DRR) are fields of research, policy, and practice relevant for multiple sectors and governance. Although they pursue complementary objectives, they have different structures and different policies [19], and thus are characterised by different norms and value systems. Such formative differences can hardly be bridged by legislation or financial incentives alone. They need effective communication and collaboration to overcome the lack of a common conceptual apparatus and shared objectives, and support the establishment of common practices and the development of synergies between them.
The SNA presented in this paper aimed at improving the understanding of the way that 35 key CCA and DRR actors active in Europe interact with each other. Communication and collaboration were found to exist both within and between the CCA and DRR groups. In relative terms, communication appears to be stronger within groups, while collaboration is also relatively relevant between them. The existence of two "communities" corresponding to actors involved mainly in CCA or in DRR activities was confirmed both in the SNA and follow-up interviews. The allocation of actors to the two groups was not rigid and variations could be observed depending on the applied algorithms and the considered indicators. Overall actors tied with DRR related connections were more prone to clustering, i.e., they had a recognisable pattern of connections, while CCA and mixed related connections were more diverse, without an obvious clustering affinity.
The exercise identified actors with a strong role in the studied network. The European Climate Adaptation Platform, Climate-ADAPT, had the highest value for various SNA metrics, reflecting its popularity in the network and its potential for enhancing interactions among its actors. Other EU institutions, research institutes, and international bodies had prominent roles in the considered network, along with some international networks, platforms, and representatives of the private sector.
Although SNA revealed overall a well-connected network, in-depth interviews described communication between both communities as insufficient, while emerging collaboration was traced back to flagship projects and initiatives to connect both domains, which are still rare. Follow-up interviews also pointed at a possible perception bias related to actors who do not identify with a specific group (CCA or DRR) but are an integral part of it from the perspective of their counterpart. This suggests that interactions between CCA and DRR communities may be more frequent in reality than indicated in the SNA. As in most complex systems, several factors may hinder constructive interactions between CCA and DRR. It is important to understand such dynamics and to draw tentative recommendations that would support communication and collaboration interactions between the CCA and DRR communities, and ultimately enhance the alignment of the two domains and their relevant initiatives.

Acknowledgments:
The authors would like to thank all the experts who participated in the survey and the interviews for their valuable inputs, as well as the four anonymous reviewers for their comments and suggestions. We also like to thank Peter Walton of Environmental Change Institute of Oxford University for a thorough native speaker review.

Conflicts of Interest:
The authors declare no conflict of interest. Table A1. List of actors invited to participate in the online SNA survey.

No.
Actor's Name Actor's Acronym

Interview guidelines
Questionnaire on the results of the Social Network Analysis of the PLACARD Project Your institution turned out to play an important role in the analyzed network. In order to better understand and explain the reasons behind the obtained outcomes we would like to ask you a few additional questions. Thank you in advance for your cooperation.  Table A1) correspond to your perception?
(b) If there are substantial differences between the SNA outcomes and your answers related to expected outcomes: are you positively/ negatively surprised? Why?
Q2. Climate-ADAPT was best rated in terms of its ability to create a bridge between CCA & DRR community. What could be the possible reason for a particularly strong position of the Climate-ADAPT platform in this respect, taking into account that Climate-ADAPT is a platform resulting from a partnership between DG RTG, DG CLIMA, DG JRC and EEA?
Q3. In your opinion, which organizational characteristics and external factors are likely to influence an actor's openness towards the other community in terms of network connections? Is it the organization's mission/vision including interdisciplinary work approach, experience in inter-organizational cooperation, leadership with excellent organizational and communication skills, organizational resources (e.g., finance, time, physical space, materials, equipment, and skilled personnel) or joint funding, legal requirements to conduct joint projects? Or maybe something else? Please indicate which factor(s) play(s) most important role in your opinion (max. 3) Q4. SNA results show that DRR actors tend to pertain to a single cluster, while CCA actors are generally more spread (See Figure 1). What could be the possible reason for that?
Q5. In terms of type of interactions, we found out that collaboration in our network is most often related to BOTH CCA & DRR, while communication process in many cases tackles JUST ONE of the two areas. What could be the possible explanation for this?   Providing access and sharing information on the following topics: Expected CC in Europe, current and future vulnerability of regions and sectors EU, national and transnational adaptation strategies and actions, adaptation case studies and potential adaptation options, tools that support adaptation planning.

2
Climate Adaptation Services (CAS) ( . . . ) a non-profit organisation that provides user-centred visualisation tools to anticipate the effects of climate change. CAS supports governments, policy makers, and professionals in gaining an understanding of (local) climate impacts and provides methods and tools to support the process of spatial adaptation.
Making climate information accessible and understandable to practitioners: Visualisation of climate change impacts, interpretation of climate scenarios.

Climate-Knowledge and Innovation Community (Climate-KIC)
Europe's largest public-private innovation partnership focused on climate change, consisting of dynamic companies, the best academic institutions, and the public sector.' Its mission is to create zero-emission, resilient society. KIC was established in 2001 by European Institute of Innovation and Technology (EIT -EU body).
Integration of education, entrepreneurship, and innovation resulting in connected, creative transformation of knowledge and ideas into economically viable products or services in the following areas: Urban transitions, sustainable production systems, decision metrics and finance, sustainable land use.

Climate Knowledge Brokers Group (CKB)
An alliance of leading global, regional, and national websites specialised in climate and development information. Its focus is on online initiatives that play an explicit knowledge brokerage role, rather than being simply institutional websites Developing common tools for an efficient sharing of and searching for climate data, initiating joint projects to connect different websites, test out innovations and generate new thinking on how to best meet user needs, supporting peer learning, building capacity by offering hands-on advice to new initiatives being set up, providing leadership to create a focal point and a voice for the community as a whole.
Climate Research Translation of complex scientific data and analyses into usable information, ( . . . ) science-based advice for DRM policies, as well as timely and reliable scientific-based analyses for emergency preparedness and coordinated response activities. It brings together existing initiatives in which science and innovative practices contribute to the management of disaster risks DRR

11
Environment and Climate Regional Accession Network -ECRAN EU-financed and EC-managed programme that supports cooperation between EU and prospective EU candidate countries in the area of environmental and climate protection as well as the alignment of their policies and actions to the EU standards. The work of ECRAN is conducted in three WPs related to environment, climate action, and cross-cutting issues (such as enforcement and compliance or public participation).
Facilitation of exchange of information and promotion of best practices related to the EU accession and corresponding environmental requirements. In particular, the Adaptation Working Group aims to encourage development and implementation of adaptation policies in the beneficiary countries, to make economic sectors incl. infrastructure 'climate proof', to improve adaptation decision making process and connect it to WeAdapt.
Environmental protection, CCM, CCA Facilitating strategic alliances, advocacy, exchange of experiences and best practices, organisation of campaigns and programmes, provision of tools and technical trainings to enable sustainable growth.
(Urban) Sustainability User-driven development and provision of knowledge for understanding the climate, climate change and its impacts, as well as guidance in its use. Industry exposure and event loss data, and an associated industry loss index service for windstorm in Europe, flood in UK, earthquake and flood in Italy, and earthquake and flood in Turkey.

24
Platform on Natural Hazards of the Alpine Convention (PLANALP) PLANALP 'was set up as part of Alpine Convention to develop common strategies designed to prevent natural hazards in the Alps as well as to exchange on adaptation strategies Formulation of strategic concepts on integrated risk management against natural hazards and the coordinated implementation of subsequent measures, knowledge transfer and exchange of best practices, implementation of flood (risk) management plans, propagating the use of standards for natural hazards and risk.

25
PreventionWeb 'participatory web platform for the disaster risk reduction community', created by UNISDR.
facilitation of an understanding of the subject of disaster risk reduction (DRR) and the work of professionals in this area by providing current news and views on the topic, and tools for exchange and collaboration.' Working Community of the Pyrenees is a consortium of 8 territorial governments (7 sub national governments and one national government) launched in 2010 the Pyrenees Climate Change Observatory. The OPCC's goal is to monitor and understand the climate evolution in the Pyrenees, with the aim of becoming less vulnerable to the impacts of climate change, and adapting to its effects by defining appropriate adaptation strategies for socio-economic sectors and the most fragile natural areas.

DRR
The objectives of the OPCC are the following: (1) Gathering the existing knowledge on impact of CC in the Pyrenees and identifying knowledge gaps, (2) analysing vulnerability in the region and its socio-economic impact, (3) provision of recommendations concerning adaptation to climate change to ensure sustainable development, (4) education of population, (5) propagation of adaptation and observation measures taken in the Pyrenees.
Sustainability, CCA Re-insurance and risk expertise provider established in 1863. Its goal is to make the world a more resilient place.
Operation of databases on disasters and loss and damage, developing models to predict future events and their potential impact, calculation of the corresponding risk premium Insurance

30
Inter-agency Working Group on CC and DRR This group was established to share information between the disaster risk reduction and climate change communities. It has been supported by UNISDR and co-chaired by UNDP and WMO.
Gathering and sharing good practices in climate risk-reduction, providing policy guidance to UNFCCC processes on the post-2012 climate change regime and developing methods for reducing the carbon footprint of disaster risk reduction activities.

31
UNISDR Europe UNISDR regional office for Europe (...) Ensuring disaster risk reduction (DRR) is applied to climate change adaptation, increasing investments for DRR, building disaster-resilient cities, schools, and hospitals, and strengthening the international system for DRR.

UNISDR STAG
The purpose of the STAG is to provide 'technical advice and support in the formulation and implementation of activities carried out by the disaster risk reduction community.' The work of the STAG encompasses all aspects of the scientific and technical dimensions of risk reduction, with particular emphasis on the needs of developing countries. STAG consists of up to 20 representatives of the institutions involved in the Scientific and Technical Partnership and a couple of individual experts.
Improving policymaking and practice by the means of scientific research, interdisciplinary multi-hazard research, translation of science into practice, bringing together national and regional DRR platforms and networks. Collaborative online 'open space' on climate adaptation issues and synergies with mitigation, which allows practitioners, researchers, and policy makers to access credible, high-quality information and to share experiences and lessons learnt.

DRR
Facilitation of learning, exchange, collaboration, and knowledge integration with the aim of building a professional community of research and practice on adaptation issues, developing policy-relevant tools and guidance for adaptation planning and decision-making. CCA