Real-World Application of Ego-Network Analysis to Evaluate Environmental Management Structures
In a world in a constant need for development, preserving biodiversity is a daunting task for both governments and NGOs. The centerpiece of successful biodiversity conservation is ensuring cooperation among countless actors involved in the management of protected areas. Social network analysis is a suitable tool for securing essential information for interactions during the management process. To contribute to the debate in the field of governance of protected areas, we illustrate an approach in investigating management of Natura 2000 sites, by considering two real-world management settings in Romania. We evaluate the characteristics of two ego-networks established for the management of two European Union Natura 2000 protected areas in Romania Iron Gates Natural Park administrated by public body owned by state and Lower Siret Floodplain administrated by a regional NGO. The networks were created around administrative bodies of protected area (ego), and include actors directly connected to the ego. After evaluating the most common ego-network metrics that demonstrate the characteristics of each network, we analyzed the strong ties by using Simmelian ties within protected areas management ego-networks and clustered the embedded links in Girvan–Newman groups. The findings suggest that the ego (administrative bodies of protected area) has a critical role in bridging other management actors. The paper tries to identify models of management control by comparing two ego-networks and showing how well connected the administrative bodies of protected areas are. The study provides insights regarding several means to improve the cooperation of environmental conservation in Natura 2000 areas.
KeywordsGovernance Natura 2000 network Simmelian ties Clusters Ego-networks Romania
Natura 2000 was established at European Union level as a network of protected areas which includes a representative sample of wildlife and natural habitats of community interest. The selection, designation, and management of Natura 2000 sites are governed by two European Union Directives: Habitats Directive for Sites of Community Interest and Birds Directive for Special Protection Areas [1, 2]. The expansion of Natura 2000 network in EU countries is nearly complete; however, obstacles remain in the implementation and enforcement of the both directives, mostly due to the lack of streamlined management models . The management of these protected sites must consider that a Natura 2000 site is primarily a tool for conservation of social–ecological systems and not focused solely on strictly ecological protection. Given that it is such a complex social–ecological arena, collaborative partnerships may be the key to succeed in implementing conservation goals [4, 5].
Most of the scientific literature focused on Natura 2000 analyzes environmental issues without considering co-management as a key strategy to improve environmental protection and conservation [6, 7]. Considering that Natura 2000 should be focused on human–nature relationships , there is the necessity to investigate means to improve and support the effective management of these sites , including the investigation of different models of collaboration in the management processes.
While social network literature is extensive regarding global properties of a network , relatively few studies discuss the importance of this field in correlating the benefits and constrains within a governance arena in a complex framework [11, 12], such as Natura 2000 . Identifying the actors that are involved in management networks by analyzing network properties at actors level (such as degree and betweenness centrality metrics) may inform about key players; however, the validity of conclusions depends on the quality of data . Management networks are also analyzed at network-level metrics such as network density and clustering informing about interactions of organizations in the entire network . Compared to a full network, analyzing an ego-network presumes the investigation of the personal network, namely, the first order zone or first neighborhood . This analysis focused on the investigation of a particular management actor (ego) and all other organizations (alters) connected to the ego can reveal the importance of the collaborative management of protected areas.
Simmelian ties are usually used to extract the interaction structure of each network, in our case the interaction of the management actors, making them easier to visualize and analyze . Borgatti et al.  defines a Simmelian tie as a reciprocally connected pair with mutual ties to third parties and hence, it is an edge embedded in a clique or triple. Simmelian ties help to identify actors with shared interests and common goals by mitigating competition . By providing a conditional triadic interpretation of both network structures helps to get an insight to the social structures of the protected area management networks by highlighting the most redundant and strong ties between the protected area management actors .
Romania, a member of the European Union since 2007, designated over 600 Natura 2000 sites, covering more than 22% of its terrestrial surface [20, 21]. However, at national level, a proper implementation of the Habitats and Birds Directives’ provisions is still lacking, the main issues being related to deficiency of adaptive conservation planning and management [9, 22].
Given the social focus of Natura 2000 protected areas and large size of sites, the management of Romanian sites is a task managed by many actors including local, regional, and national environmental protection authorities, local and regional administrative authorities, custodians of Natura 2000 sites, local inhabitants, industry, advisory boards, and other local, regional, or national partners (e.g., NGOs) . Romanian protected areas management system (i.e., delegated management to organizations governed by different jurisdictions ) might be a good case study to analyze the interaction of protected areas administrators with other organizations involved in management.
To gain insight into how protected areas network relational structure develops around two different types of leading management organizations, our study focuses on two protected sites, namely, Iron Gates Natural Park (administrated by a state-owned enterprise with a public body statute) and Lower Siret Floodplain Natura 2000 (administrated by an NGO) . NGOs are known to have a significant impact on environmental conservation actions and in helping to integrate environmental objectives into policy and practice , hence, we hypothesize that the main characteristics of the two analyzed networks differ because they are driven by two different types of egos (public body versus NGO). Furthermore, we hypothesize that the NGO ego-net would be denser than the one driven by a public body, with consequences in the efficiency of any management process.
Obtaining data on the different structural characteristics of each ego-network will help to understand the different patterns of management in terms of cooperation. Furthermore, this paper focuses on the evaluation of protected areas management by considering social patterns of cooperation and highlighting best practices to further contribute to biodiversity conservation .
This chapter represents an extension of our previous conference paper , in which we briefly described the benefits of characterizing protected areas management using ego-networks. The data-sets used for the ego-network analyses were previously used to describe the current state of the governance network structure .
2.1 Study Areas
In 2007, Iron Gates Natural Park became part of Natura 2000 due to its richness of biodiversity of European importance. The park overlaps two Special Protection Areas (SPAs) and one Site of Community Importance (SCIs) . Lower Siret Floodplain Natura 2000 protects 22 bird species of European interest that are found in Annex I of the Birds Directive, representing a significant hot spot for migratory birds .
2.2 Research Design and Data Collection
We considered as members of the management networks the stakeholders involved in or affected by decisions related with analyzed Natura 2000 sites (nodes) and the links between them (edges or links).
Initially, we identified potential members of the two networks by analyzing the management plans of the two protected areas. Then, between April and June 2016 we surveyed the representatives of 60 organizations from Iron Gates Natural Park and 65 organizations from Lower Siret Floodplain. The survey gathered data on the collaborative relationships established during the management activities. We found that the Iron Gates Natural Park and Lower Siret Floodplain management networks include 99 and 120 organizations, respectively. The difference between the number of actors is mostly because Iron Gates Natural Park is located in two counties (Mehedinti and Caras-Severin), while Lower Siret Floodplain in three counties (Vrancea, Galati, and Braila) .
We considered ego-nets as appropriate for our analyses because of their constrained and simpler structure . To reduce the size of the networks, and better identify the potential differences between the two case studies, we further analyzed the structures created around the egos.
Conclusively, our study brings into discussion and analyzes the ego-networks composed of the most important players in the protected areas management.
3 Concepts and Methodology
Comparing the ego-networks structural metrics lead by different types of organization (i.e., NGO versus public body);
- 2.Correlating the indegree and betweenness values of the networks in three case scenarios:
before extracting the ego-networks;
after extracting the ego-networks of the main management organizations;
after removing the ego from both networks.
Analyzing the consequences after removing the ego by applying the Simmelian ties algorithm and Girvan–Newman clustering.
Interpretation of analyzed ego-networks metrics
Interpretation in our analysis
Number of management actors directly connected with the protected area administration organization 
Number of ties
Number of links among all management actors in the ego-network 
Number of connections divided by the number of pairs, namely, the percentage of all possible links in each ego-network 
A weak component is the largest number of management actors that are connected, disregarding the direction of the link 
The percentage of all geodesic paths from neighbor to neighbor that pass through the protected area administration organization 
Compares the actual betweenness of the protected area administration organization to the maximum possible betweenness in neighborhood of the size and connectivity of ego’s. The maximum value for betweenness would be achieved where ego is the center of a “star” network, that is, no neighbors communicate directly with one another, and all directed communications between pairs of neighbors go through ego 
the complete management network;
the ego-network of the main administrative management organization;
the remaining management actors after removing the ego.
We carried out this three-step analysis to demonstrate how actors migrate and establish a distinct position within the network when changing the size of the network by removing the egos.
The final step focuses on the consequences of removing the ego by analyzing the role of Simmelian ties in the networks and clustering using Girvan–Newman algorithm, which describes a hierarchical method regularly used to detect communities in complex systems based on edges betweenness [25, 35]. Although slower than other clustering techniques, we choose to use Girvan–Newman algorithm to investigate our ego-networks because it is well documented and often used in popular network analysis programs , thus serving well our aim to demonstrate a real-world application of ego-network analysis to differentiate environmental management structures. Giving the potential to transfer the results obtained for an ego-net to a complete network , we further extracted substructures for each case study consisting of ties that are strong and redundant . To learn what would happen with the management networks (thus, the whole management of the protected area) if the egos (i.e., protected areas administration) would disappear, we applied the Simmelian ties algorithm, using the triples method “mutual friends”  between each pair of actors, before and after removing the protected area management administration main actor (ego of each case study). Social ties embedded within triads are more stable over time, stronger, and more durable . In our case, analyzing this is important, because informational flow can be obstructed if involved management organizations are not being proactive or are acting opportunistically and avoid exchanging knowledge and information with other actors of the network .
We illustrate the results of this application using NetDraw  and VosViewer . Additionally, to identify community structure in both networks, we grouped the actors in clusters using Girvan–Newman algorithm . Typically, this hierarchical clustering is used to remove the highest betweenness edges until the network falls apart; however, our principal objective is to compare the two managements of the protected areas by removing the ego. In our paper, we reported the results of the Girvan–Newman algorithm by grouping the management actors in clusters (i.e., corresponding to maximum three counties overlapping the protected areas), without specifying how many groups shall be formed or assigning restrictions on their size .
4 Research Findings and Discussion
4.1 Structure of the Ego-Networks
Properties of the ego-network case studies
Case study 1: Iron Gates
Case study 2: Lower Siret
Initial size of the management network
Size of management EgoNet (including ego)
Number of weak components divided by sizes
Our results indicate that the ego-network of the Iron Gates Natural Park is denser with lesser number of weak components. Conversely, Lower Siret Floodplain ego-network has a higher betweenness, whereas the weighted overall graph clustering coefficient appears to be similar for both analyzed cases (see Table 2).
Even though the number of organizations participating in the management of Lower Siret Floodplain is higher than in the Iron Gates Natural Park, the ego is less connected with the other management actors from the territory. This is expected since most organizations involved in or affected by decisions related to management of analyzed protected areas are public bodies, and usually organizations with different jurisdictions (i.e., public bodies vs. NGOs have a lower tendency to cooperate, mostly due to the difference between governing norms) [39, 40].
4.2 Correlation Between the Indegree and Betweenness Values of the Networks
This confirms the potential weakness of Lower Siret Floodplain management network, where actors are connected only with protected area administration and do not communicate directly. In the case of Iron Gates Natural Park network, the communication between actors can be efficient even if the protected area administrator is not responsive. This can be interpreted as an ability of other public bodies from the respective management network to act as redundant connections for a public body protected area administration, while the NGOs administrating a protected area must communicate actor by actor to perform management activities in their area of interest, particularly in environmental protection and conservation. This analysis confirms the pattern resulted from interpreting the ego-networks properties, and most importantly, underlines that the NGOs administrating a protected area must strengthen the cooperation with key public body actors, such as environmental protection agencies and local and regional administrative authorities.
4.3 Consequences of Removing the Ego: Role of Simmelian Ties and Girvan–Newman Clustering
After removing the ego, the Lower Siret Floodplain network falls apart, because the management organizations, except the ego, have roles only at local level and do not ensure communication within the entire network, failing to ensure a flow in the management processes. We were expecting that protected areas administrated by an NGO to be better connected to other organizations than one governed by a public body [5, 41]. From this perspective, the Iron Gates Natural Park network has a more effective collaborative management, adopting the premise that the managerial organizations within a protected area should not become too dominant but recognize others as collaborative partners . This, along with enhanced communication flow, sharing responsibilities and transfer of knowledge, may be the key for successful management of protected areas.
As it can be seen in Figs. 6 and 7, by taking Simmelian ties into account after clustering the organizations, we defined three categories of network actors, this way predicting the future of a management network if ego (main management organization) would disappear. Figure 7 reinforces our prior results (structure of the ego-networks and correlation between the indegree and betweenness values of the networks) by identifying in Iron Gates Natural Park management network, stronger Simmelian bridges, the most important management actors remaining connected and thus ensuring a faster information flow and more qualitative associated management processes.
5 Conclusions and Future Work
Integrating social network and ego-network analyses within environmental management provides an overview of the actual patterns of the structure, but the results obtained for each environmental management network could not be generalized to all protected areas, the findings being beneficial for stakeholders and practitioners in the field.
The current study contributes to our understanding of the linkage between protected areas management actors. By analyzing ego-networks structural metrics, correlating the indegree and betweenness values, and applying the Simmelian ties algorithm and Girvan–Newman clustering after removing the ego, we successfully achieved a diagnosis of main social structures in the management of two Natura 2000 sites, representative for the management of protected areas in Romania, namely, Iron Gates and Lower Siret Floodplain. Using ego-network analysis, we have succeeded in obtaining significant information and contribute to the environmental management field by showing a way to facilitate more accurate and efficient management analysis .
Our results demonstrate large variations in ego-networks metrics, indicating that the ego-networks properties depend on the ability of the ego to connect to organization with different jurisdictions . In our case, the network coordinated by an NGO (Lower Siret Floodplain) is more fragmented, and such, the collaboration and information flows are strongly influenced by the existence of one key actor only. Such networks are more vulnerable to disruptions of collaborations , for example, when the management body tries to enforce more strict regulations on resource management to protect the biodiversity. Without taking in consideration the management performance, after comparing the two management networks, we could consider that Iron Gates Natural Park management network is an example of best practice in structural terms.
Examining the effect of tie strength presents an opportunity to predict what role the stakeholders play within the management of the protected areas. Additionally, using embedded ties and link strengths showed for Lower Siret Floodplain a higher potential for collapse if removing the ego from the network.
Our results about network formation and structure could be further developed by comparing network structure over time  and be extended to other protected area management, presenting informative potential for managers. Considering expanding our results to other protected areas, the level of cooperation and actions will positively change so that the conservation targets may be achieved without stimulating social conflicts [47, 48]. In line with previous findings , using social network analysis in protected area management can help involved actors to understand the potential risks of weak network components and predict arrangements that can undermine conservation efforts.
Interpreting Simmelian ties findings could play a defining role in training and forming future leaders. From this point of view, this technique could have a significant potential to establish the development of future leaders within environmental management networks.
We demonstrated that network analysis can contribute to improve protected areas management and may be a useful tool for systematic conservation planning . Our findings could be further used also to minimize the protected area network’s vulnerabilities and predict the potential for large-scale failure. From this point of view, using ego-network analysis within the protected areas management can be a starting point to adapt over time and recover after a network disruption and hence contribute to the implementation of Habitats and Birds Directive. This is why, network analysis may play a defining role in good resources management promotion [50, 51]. Nevertheless, further work is needed to create an accessible streamlined methodology so that the protected areas managers use the insights that social network analysis could provide to conservation planning.
We would like to thank Iron Gates Natural Park and Lower Siret Floodplain administrations and Iulia Viorica Miu for the data collection process, and to Edward F. Rozylowicz for proofreading the manuscript. This work was supported by a grant of the Romanian National Authority for Scientific Research, CNCS—UEFISCDI, PN-III-P4-ID-PCE-2016-0483 (http://www.uefiscdi.ro).
- 1.European Commission: Council Directive 92/43/EEC of 21 May 1992 on the conservation of natural habitats and of wild fauna and flora (Habitats Directive). Off. J. Eur. Union L. 206, 7–50 (1992)Google Scholar
- 2.European Commission Directive 2009/147/EC of the European Parliament and of the Council of 30 November 2009 on the conservation of wild birds: Off. J. Eur. Union L. 20, 7–25 (2009)Google Scholar
- 3.European Commission: Environmental impact assessment of projects: rulings of the court of justice. Report, European Commission (2013)Google Scholar
- 4.Hanspach, J., Hartel, T., Milcu, A.I., Mikulcak, F., Dorresteijn, I., Loos, J., von Wehrden, H., Kuemmerle, T., Abson, D., Kovacs-Hostyanszki, A., Baldi, A., Fischer, J.: A holistic approach to studying social-ecological systems and its application to southern Transylvania. Ecol. Soc. 19(4), art32 (2014). https://doi.org/10.5751/Es-06915-190432
- 6.Blicharska, M., Orlikowska, E.H., Roberge, J.M., Grodzinska-Jurczak, M.: Contribution of social science to large scale biodiversity conservation: a review of research about the Natura 2000 network. Biol. Conserv. 199, 110–122 (2016). https://doi.org/10.1016/j.biocon.2016.05.007 CrossRefGoogle Scholar
- 10.Borgatti, S.P., Everett, M.G., Johnson, J.C.: Analyzing Social Networks, 2nd edn. SAGE, London (2017)Google Scholar
- 11.Bodin, O., Robins, G., McAllister, R.R.J., Guerrero, A.M., Crona, B., Tengo, M., Lubell, M.: Theorizing benefits and constraints in collaborative environmental governance: a transdisciplinary social-ecological network approach for empirical investigations. Ecol. Soc. 21(1), art40 (2016). https://doi.org/10.5751/Es-08368-210140
- 13.Rey, V., Groza, O., Patroescu, M., Ianos, I.: Atlas de la Roumanie: dynamiques du territoire. Reclus. La Documentation française (2007)Google Scholar
- 17.Nick, B., Lee, C., Cunningham, P., Brandes, U.: Simmelian backbones: amplifying hidden homophily in Facebook networks. In: 2013 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), pp. 531–538 (2013). https://doi.org/10.1145/2492517.2492569
- 18.Borgatti, S.P., Everett, M.G., Freeman, L.C.: Ucinet for Windows: Software for Social Network Analysis. Analytic Technologies, Harvard (2002)Google Scholar
- 20.Manolache, S., Ciocanea, C.M., Rozylowicz, L., Nita, A.: Natura 2000 in Romania - a decade of governance challenges. Eur. J. Geogr. 8, 24–34 (2016)Google Scholar
- 22.Nita, M.R., Niculae, I.M., Vanau, G.O.: Integrating spatial planning of protected areas and transportation infrastructures. In: Using Decision Support Systems for Transportation Planning Efficiency, pp. 311–329. Engineering Science Reference, Hershey (2016)Google Scholar
- 25.Nita, A., Manolache, S., Ciocanea, C., Rozylowicz, L.: Characterizing protected areas management using ego-networks. In: 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, pp. 642–643 (2017). https://doi.org/10.1145/3110025.3110079
- 27.Iron Gates Natural Park Administration: Planul de management al Parcului Natural Porţile de Fier. Report, RNP Romsilva-Administratia Parcului Natural Portile de Fier (2013)Google Scholar
- 28.Association for Biodiversity Conservation: Planul de management al ROSPA0071 Lunca Siretului Inferior si al ariilor naturale protejate suprapuse. Report, ACDB (2015)Google Scholar
- 29.Chang, A.X.: Analysis of Email Ego Networks. An exploratory study of ego networks in an email network. Snap.Stanford.Edu (2010)Google Scholar
- 30.Harrigan, N., Achananuparp, P., Lim, E.P.: Influentials, novelty, and social contagion. The viral power of average friends, close communities, and old news. Soc. Netw. 34(4), 470–480 (2012). https://doi.org/10.1016/j.socnet.2012.02.005
- 32.Hertzberg, V.S., Baumgardner, J., Mehta, C.C., Elon, L.K., Cotsonis, G., Lowery-North, D.W.: Contact networks in the emergency department: effects of time, environment, patient characteristics, and staff role. Soc. Netw. 48, 181–191 (2017). https://doi.org/10.1016/j.socnet.2016.08.005 CrossRefGoogle Scholar
- 34.Hanneman, R., Riddle, M.: Introduction to Social Network Methods. University of California, Riverside (2005)Google Scholar
- 39.Ostrom, E.: Understanding Institutional Diversity. Princeton University Press, New Jersey (2005)Google Scholar
- 42.Fliervoet, J.M., Geerling, G.W., Mostert, E., Smits, A.J.: Analyzing collaborative governance through social network analysis: a case study of river management along the Waal river in The Netherlands. Environ. Manag. 57(2), 355–67 (2016). https://doi.org/10.1007/s00267-015-0606-x CrossRefGoogle Scholar
- 43.Li, C., Lin, S.: Egocentric information abstraction for heterogeneous social networks. In: 2009 International Conference on Advances in Social Network Analysis and Mining, pp. 255–260 (2009). https://doi.org/10.1109/ASONAM.2009.38
- 44.Ionita, A., Stanciu, E.: Protected area governance in Eastern Europe. Report, BfN-Skripten 360 (2015)Google Scholar