Strengths and Vulnerabilities of Australian Networks for Conservation of Threatened Birds

Strengths and vulnerabilities of Australian networks for conservation of threatened birds

T IM Q. HOLMES,BRIAN W. HEAD,HUGH P. POSSINGHAM and S TEPHEN T. GARNETT

Abstract We analysed the supportive social networks asso- making actors within an organizational framework trying ciated with the conservation of six threatened Australian to solve collective problems. The participants in an action bird taxa, in one of the first network analyses of threatened situation (Ostrom, ) for a threatened species may in- species conservation programmes. Each example showed clude those actively involved in the conservation of the contrasting vulnerabilities. The Alligator Rivers yellow bird or those with an interest in, or concern about, the chat Epthianura crocea tunneyi had the smallest social net- bird or a related issue. They may be individuals, or represen- work and no real action was supported. For the Capricorn tatives of NGOs, community groups, land owners, govern- yellow chat Epthianura crocea macgregori the network was ments or corporate actors, interacting for a range of reasons, centred on one knowledgeable and committed actor. The including obtaining or disseminating information, solving orange-bellied parrot Neophema chrysogaster had a strongly problems, negotiating, making decisions, collaborating connected recovery team but gaps in the overall network and seeking physical resources (Ostrom, ; Jepson could limit communication. The recovery teams for the et al., ). The key components of these action situations swift parrot Lathamus discolor and Baudin’s black-cockatoo include key participants, interorganizational collaborations Calyptorhynchus baudinii had strong links among most and social networks. Here, we focus on the social networks stakeholders but had weak ties to the timber industry and of key participants and seek to determine how such net- orchardists, respectively, limiting their capacity to manage works may influence the effective management of biodiver- threatening processes. Carnaby’s black cockatoo Calyptor- sity protection schemes for threatened birds. To our hynchus latirostris seemed to have the most effective social knowledge this is one of the first analyses of social networks network of any of the taxa studied but may be vulnerable to concerned with the conservation of threatened species (see skill shortages. In each case the network analysis pointed to also Ainsworth et al., ; Jepson, ). gaps that could be filled to enhance the conservation effort, In the context of conservation policy and management a and highlighted the importance of recovery teams. The re- social network comprises a set of actors with socially mean- search suggests that formal network analysis could assist in ingful ties that represent relationships such as those arising the design of more effective support mechanisms for the from shared decision making, seeking or providing advice, conservation of threatened species. or borrowing or lending resources. Social networks can be useful as a basis for establishing informal collaborations Keywords Connectedness, governance, recovery teams, and for working across critical functional, hierarchical or social network analysis, threatened species geographical boundaries (Cross et al., a,b). Links among actors at multiple levels have also been found to result in greater capacity for monitoring, improved ecological Introduction understanding and promotion of appropriate incentives (e.g. power sharing, distribution of economic benefits; uch has been written about the biology of Australian Armitage et al., ). Increasingly networks are being Mthreatened birds; however, less information is avail- viewed as an evolving form of governance supplementing able about the social and organizational processes involved formal and hierarchical approaches (Stoker, ; Keast in their management. These social processes primarily refer et al., ). However, like all forms of governance, there to the cooperative spaces where participants strive to achieve is variety among networks in terms of their processes, mem- a common purpose (in this case, the conservation of the bership, resources and legitimacy (Adger et al., ; Koliba threatened species). The key participants are the decision- et al., ). As governance instruments, all networks involve some form of trade-off (Alexander et al., ), so there is no

TIM Q. HOLMES and HUGH P. POSSINGHAM Centre for Biodiversity and universal ideal, but analysis of networks can help make such Conservation Science, The University of Queensland, St Lucia, Australia trade-offs explicit and facilitate assessment of how they are

BRIAN W. HEAD Institute of Social Science Research, The University of influencing the governance issues of concern. Queensland, St Lucia, Australia Social network analysis has been applied to both the STEPHEN T. GARNETT (Corresponding author) Research Institute for the physical and social sciences, with examples including fields Environment and Livelihoods, Charles Darwin University, Darwin, Northern Territory 0909, Australia. Email [email protected] as diverse as the spread of happiness (Fowler & Christakis,       ), the study of animal behaviour (Wey et al., ), and Received March . Revision requested April .  Accepted  April . First published online  November . the internet as a social network (Otte & Rousseau, ).

Social network research has become more widely under- bird taxa in Australia. The analysis helped identify the taken, with at least five peer-reviewed journals (Social strength and nature of organizational connections in the Networks, Computational Social Networks, Social Network political landscape by quantifying the intensity of relation- Analysis and Mining, International Journal of Social ships and interactions among the actors (Holland, ). Network Mining, Journal of Social Structure) now dedicated Some of the key assumptions and underlying concepts of to publishing social network research. Social network ana- our analysis include the following (Wasserman & Faust, lysis identifies patterns of relationship among people and ): relationships among actors are important, actors organizations, and can represent them both visually and and their actions are seen as interdependent rather than in- mathematically. In general, it has proved useful for provid- dependent, relational ties between actors are channels for ing a formal mechanism for representation, measurement, the transfer of resources (either material or nonmaterial), and modelling of relational structures. Although there is al- and network models help us to conceptualize a range of ways a risk that methods may be misapplied or misinter- weak to robust patterns of relations among actors. preted, resulting in inappropriate conclusions, the risk can Based on the analyses provided by Bodin et al. (), it be minimized through careful choice of methods and appli- is possible to investigate and measure several relational fac- cations in network analysis (Butts, ). In recent years so- tors that may have impacts on learning, leadership and trust, cial network analysis has been employed by natural resource which are relevant to the contributions of social networks management and conservation biology researchers to im- for managing natural resources such as threatened species. prove understanding of how actors collaborate and coordin- These characteristics may change or evolve over time, and ate management efforts (Tompkins & Adger, ; Bodin some factors are likely to be more significant for certain con- et al., ; Lauber et al., ; Woodward, ; Chilvers texts than others. The analysis below adapts the categories & Evans, ; Crona & Hubacek, ; Lukasiewicz et al., provided by Bodin et al. () as follows: ). This work has also recognized the importance of sta- . Centrality: the number of links to an individual actor. keholders as influential environmental decision makers This can affect the capacity of a group to experiment (Prell et al., ) and has shown that social network ana- with new ideas or coordinate actions. lysis can make valuable contributions to conservation out- . Density: the number of links divided by the number of comes as a tool for exploring the relationships between actors in the network. This is useful for understanding stakeholder network structures and the implementation of the variety of skills in a group and whether it will be af- conservation actions (Vance-Borland & Holley, ). fected by the loss of individuals. For the management of Australia’s threatened birds, rele- . Reachability: when there are connections between two vant social networks contain stakeholders within designated actors regardless of how many other actors fall between management areas and can be used to mobilize and main- them. This can affect communication within a group. tain co-management of individual threatened species. . Betweenness: the percentage of all geodesic paths from Although the effectiveness of recovery plans has sometimes neighbour to neighbour that pass through an actor. been questioned (Bottrill et al., ), recovery teams, which This can help in understanding the diversity of skills are associated with some, but not all, recovery plans, have and levels of trust within a group. been influential within these networks (Holmes, ). . Subgroups: actor groups that can be identified among the Species recovery teams are collaborations of stakeholders full component of individuals within the network. This who coordinate the implementation of the recovery plans. can indicate the extent to which alternative views are Key roles of recovery teams include coordination, commu- likely to be available to generate innovation. nication (including to their broader networks) and inclusion of stakeholders (Holmes, ). The objectives of this study These six factors, all of which have relevance to various as- are to describe the networks established among those in- pects of threatened species governance, have been combined volved in the conservation of threatened birds, how they in Table  as a framework for assessing the key attributes of contribute to solving the problems faced in threatened spe- the six species conservation networks that are the focus of cies conservation, and the value that can be derived from this study. analysing them. Case study bird taxa Methods We examined the social networks for six threatened birds Network analysis is undertaken to understand the pattern of (the Alligator Rivers yellow chat Epthianura crocea tunneyi, relations among individuals and organizations with a direct the Capricorn yellow chat Epthianura crocea macgregori, interest in a topic of common interest (Wasserman & Faust, the orange-bellied parrot Neophema chrysogaster, the swift ; Bodin & Crona, ). Here we have analysed the net- parrot Lathamus discolor, Carnaby’s black-cockatoo works associated with the conservation of six threatened Calyptorhynchus latirostris, and Baudin’s black-cockatoo

TABLE 1 Attributes potentially influencing the effectiveness of social networks for the management of natural resources, and related metrics (derived from Bodin et al., ).

Attribute Network metric characteristics Collective memory/experiences: commonality in history strengthens Density: high; many links to others in the network network cohesion in times of change & uncertainty. Reachability: high; access to many individuals Diversity: variation in types of actor & of knowledge within the network Density: low; a dispersed network can increase the variety of & potential for innovation is obtained from a degree of separation be- skills available tween groups. It is also important for fostering the resilience needed to Betweenness/subgroups: low; groups within a network need a adapt to unexpected change. certain degree of separation to facilitate innovation Redundancy, or the extent to which links are replicated, helps buffer a Density: high; with many links the loss of single actors is less network against loss of individuals. disruptive Betweenness: low; low levels of betweenness of single actors reduces the vulnerability of a network to fragmentation Learning: knowledge can be increased & improved continually if there Centrality: low; decentralized management encourages experi- are strong ties & access to many actors to receive & disseminate mentation & experiential learning information. Reachability: high; access to many actors increases knowledge pool Betweenness: low; learning is facilitated by close links within subgroups to transfer knowledge Adaptive capacity: new challenges or knowledge require a timely re- Centrality: high; concentration of links increases capacity for sponse for networks to be effective. coordination of actions Density: low; actors with fewer links have greater political flexibility & are subject to lower levels of peer pressure Reachability: high; collective action requires multiple actors to collaborate Trust: cooperation is facilitated by high levels of mutual confidence Density: high; many links foster feelings of belonging & group among actors. identity Betweenness/subgroups: high; low levels of separation among groups can assist the development of trust

Calyptorhynchus baudinii), taking into account the follow- (Armstrong, ). It is categorized as Endangered under ing important contextual characteristics: type of organiza- both the IUCN criteria (Garnett et al., ) and the tion involved in threatened bird conservation; programme Environment Protection and Biodiversity Conservation structure; number of partners and participants; habitat/eco- Act  (Department of the Environment, ). The system; threats; land tenure; scale; and jurisdictional over- main identified threats to the Alligator Rivers yellow chat lap. In selecting the six taxa we chose three sets of pairs are habitat degradation by introduced herbivores, such as with similar attributes: () two subspecies of the same spe- cattle, feral pigs and water buffalo, invasive plants and salt- cies that have similar ecology and are localized in distribu- water intrusion; inappropriate fire regimes; and predation tion but separated geographically, and in separate by feral cats (Commonwealth Threatened Species jurisdictions; () two distinct species with different ecologies Scientific Committee, ). Coordination of conservation that both migrate and occur in multiple jurisdictions; and management for this subspecies has been left to individuals () two closely related species of similar appearance that within the Northern Territory’s Department of Land have overlapping distributions under a single jurisdiction. Resource Management and the Parks and Wildlife The study of these paired examples facilitated a more Commission as well as the management authorities of nuanced exploration of the interplay between ecological, Federally controlled Kakadu National Park. At the time of jurisdictional and social factors underlying the role and ef- the research there was no conservation activity directed spe- fectiveness of networks supporting the recovery teams. cifically at the subspecies. The Alligator Rivers yellow chat is a subspecies of the en- The Capricorn yellow chat is another subspecies of the demic yellow chat (Keast, ), a small insectivorous bird yellow chat (Keast, ), with a localized distribution in recorded from river catchments east of Darwin in the the Fitzroy River Delta near Rockhampton, central Northern Territory, where it is known from only a small Queensland, and several sites adjacent to Broad Sound, number of sites (Fig. . in Schodde & Mason, ;  km to the north, where it inhabits seasonally inundated Armstrong, ). It inhabits seasonally inundated alluvial marine plain wetlands (Houston & Melzer, ). It is cate- floodplains that support a sparse cover of grasses, herbs and gorized as Endangered under the IUCN criteria (Garnett sedges, and stands of mangrove around tidal channels et al., ), and Critically Endangered under the

Environment Protection and Biodiversity Conservation Act mainland south-eastern Australia for the winter, primarily  (Department of the Environment, ). The Capri- to Victoria and New South Wales (Swift Parrot Recovery corn yellow chat is potentially threatened by habitat degrad- Team, ). It was categorized at the time as Endangered ation caused by grazing, trampling and digging by under both the IUCN criteria (BirdLife International, mammalian herbivores, invasion by exotic pasture grasses, a) and the Environment Protection and Biodiversity alteration of natural water flows, expansion of industrial op- Conservation Act  (Department of the Environment, erations, and wildfire (Houston & Melzer, ). There was ). The greatest threat to survival of the swift parrot is a loose collection of mainly non-government actors who habitat loss through land clearing for plantation develop- had a common interest in the subspecies and its habitat ment and intensive native forest silviculture (Saunders & and who undertook population monitoring, surveys and re- Tzaros, ), and predation at the nest by the sugar glider search. Within this group was an ecologist who championed Petaurus breviceps, a native Australian marsupial intro- the implementation of the recovery plan for the taxon. This duced to Tasmania (Stojanovic et al., ). The Swift person primarily focused on ecological research and was the Parrot Recovery Team was established in  and has  primary contact for issues relating to the subspecies. members. This team has had the same leader since . The remaining four species all had recovery plans, which Carnaby’s black-cockatoo is a large black cockatoo with identified necessary research and management actions to white tail panels and is endemic to south-west Western stop the decline of a threatened species, support its recovery Australia. It breeds primarily in woodlands and scrubs in and enhance its chance of long-term survival in the wild the semiarid interior and moves to the coastal plain in winter (Department of Sustainability, Environment, Water, Popu- (Saunders, ). It is categorized as Endangered under both lation and Communities, ). For each of these recovery the IUCN criteria (BirdLife International, ) and the En- plans there were recovery teams that coordinated imple- vironment Protection and Biodiversity Conservation Act mentation of the plans.  (Department of the Environment, ). The decline The orange-bellied parrot is a small, largely green, parrot of Carnaby’s black-cockatoo is attributed to loss of habitat of coastal habitats, including eucalypt forest (for breeding), as a result of clearing or degradation, competition for nest salt-marshes, coastal dunes, pastures, shrublands, estuaries, sites, loss of individuals as a result of illegal activities (shoot- islands, beaches and moorlands. It breeds in south-west ing to protect crops and taking birds for avicultural markets), Tasmania and migrates to south-eastern South Australia collisions with motor vehicles, and disease (Department of and Victoria for the winter (Orange-bellied Parrot Recovery Environment and Conservation, ). The Carnaby’s Team, ). It is categorized as Critically Endangered Black-cockatoo Recovery Team was established in .It under both the IUCN criteria (BirdLife International, had  members and had the same leader since . b) and the Environment Protection and Biodiversity Baudin’s black-cockatoo is another large white-tailed Conservation Act  (Department of the Environment, black cockatoo endemic to south-western Australia, where ). The reasons for the species’ decline are not clear but it is largely confined to the tall wetter forests (Campbell & it was thought at the time to be threatened by fragmentation Saunders, ). Most early accounts of white-tailed black- and degradation of over-wintering habitat by drainage, graz- cockatoos did not distinguish between Baudin’s and ing, agriculture or urban development, and inappropriate fire Carnaby’s (Higgins, ). Baudin’s black-cockatoo is cate- regimes in the breeding range (Orange-bellied Parrot gorized as Endangered under the IUCN criteria (BirdLife Recovery Team, ). The Orange-bellied Parrot Recovery International, ) and Vulnerable under the Environment Team was established in  and comprised  members. Protection and Biodiversity Conservation Act  (Depart- The role of coordinator was rotated annually between repre- ment of the Environment, ). The threats identified for sentatives of the three state governments, the jurisdictions of Baudin’s black-cockatoo are illegal shooting by orchardists, which share the species’ distribution. The team recognized the habitat loss, nest hollow shortage and competition for avail- importance of the broader network, which was acknowledged able nest hollows (Chapman, ). A Recovery Team was in its terms of reference: the team ‘recognized the important established for Baudin’s black-cockatoo in ; this team contributions and interests of this broader Network and com- also covered other threatened forest cockatoos. It comprised municated to individuals through email updates. Network  members and had the same leader since its inception. members were also informed of, and invited to, all open team meetings. The Network membership was not limited Stakeholder selection and new members were included by the team as required.’ The swift parrot is a small green and red monotypic par- For each of the six case studies the relevant social networks rot of forests and woodlands, with seasonal use of habitat in all contained stakeholders within a well-defined manage- different parts of the landscape in both its breeding and ment area, making identification of the stakeholders rela- non-breeding range. It breeds on the east and south-east tively straightforward. The social networks were identified coast of Tasmania during the summer and migrates to by focusing on the individuals directly involved in the

TABLE 2 UCINET threshold scores used to characterize network metrics.

Attribute Metric Negative Positive Collective memory/experiences Density 0.15 0.39 Reachability 79% 100% Diversity Betweenness (no. of subgroups) 4 9 Density 0.39 0.15 Redundancy Betweenness (of single actors) 53% 26% Density 0.15 0.39 Learning Betweenness (no. of subgroups) 4 9 Centralization 69% 13% Reachability 79% 100% Adaptive capacity Centralization 13% 69% Density 0.39 0.15 Reachability 79% 100% Trust Density 0.39 0.15 Betweenness (no. of subgroups) 9 4

management of the case study taxa, starting with formal re- using UCINET  (Borgatti et al., ), which is specifically covery teams, if they existed, and adding new individuals if designed for mapping and measuring the characteristics of and when they emerged (through snowball sampling). For social networks. The intensity of network metrics was char- those taxa without recovery teams, individuals actively in- acterized on the basis of Table . volved in the conservation efforts were identified from relevant literature, through contacts with government officers responsible for taxon conservation, and through snowball Results sampling. Semi-structured interviews were undertaken with the identified stakeholders, providing insights into Of the  informants who were interviewed,  were mem- stakeholder relationships and triangulation of data collected bers of recovery teams, distributed as follows across the in- in networks (Reed et al., ). dividual teams: orange-bellied parrot, %; swift parrot, %; Carnaby’s black-cockatoo, %; and Baudin’s black- cockatoo, %. This represents a reasonable proportion of Stakeholder analysis key informants given that the accuracy of identifying central nodes from randomly generated social networks declines TQH conducted interviews with stakeholders in person or predictably and monotonically with inclusion (Borgatti by telephone, and included questions involving free recall et al., ), and the non-response effect was likely to (Wasserman & Faust, ), to determine with whom the have been balanced by reciprocal nominations from other stakeholders interacted, the frequency of interactions and team members (Kossinets, ). We interviewed  indivi- the strength of relationships. The definition of what consti- duals who were not on recovery teams but were directly in- tuted an interaction was left to the interviewee but was pri- volved in the recovery of the case study taxa, providing marily focused on communication, and included face-to- additional data for collaborations and social network face meetings (especially for species for which there were analysis. recovery teams), shared field experience, telephone conver- Analysis of the potentially influential attributes of net- sations and e-mail contact. We interviewed  informants works (Table ) indicates that there is substantial variation with relevant experience. The interviews comprised directed in most characteristics across the various taxa. Comparisons but open-ended questions (Holmes, ) focusing on the of the attributes indicated that many of these factors were role of the interviewee and their colleagues in the recovery likely to have had a positive influence on the effectiveness of the case study taxon. For logistical and timing reasons of networks (Table ), but there was high variability. The some members of the networks were unavailable for inter- networks themselves varied substantially, from sparse and view. In such cases their position in the network was deter- small to large and complex (Fig. ), with each characteristic mined from other network members, thus potentially having potential strengths and weaknesses. Size alone can making the effect of such omissions negligible (Kossinets, affect some metrics (Prell, ): the Alligator Rivers yellow ). Others were excluded based on the evidently low sa- chat network has only  possible linkage pairs among its lience of their involvement. nine members, whereas there are , possible linkages The data concerning social network participation and among the  members of the Carnaby’s Black-Cockatoo linkages for each of the case study taxa were analysed Recovery Team (Table ). All networks included more

Oryx, 2017, 51(4), 673–683 © 2016 Fauna & Flora International doi:10.1017/S0030605316000454 Downloaded from https://www.cambridge.org/core. IP address: 170.106.35.229, on 25 Sep 2021 at 23:59:23, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0030605316000454 https://www.cambridge.org/core/terms Downloaded from 7 .Q omse al. et Holmes Q. T. 678 https://www.cambridge.org/core . https://doi.org/10.1017/S0030605316000454 . IPaddress: TABLE 3 Attributes of the social networks for the six study species.

Capricorn yel- Carnaby’s black- Baudin’s black-

170.106.35.229 Alligator Rivers yellow low chat cockatoo cockatoo chat Epthianura crocea Epthianura cro- Orange-bellied parrot Swift parrot Calyptorhynchus Calyptorhynchus Network variables tunneyi cea macgregori Neophema chrysogaster Lathamus discolor latirostris baudinii No. of nodes 9 15 28 23 42 24 , on Possible links between node 72 210 756 506 1722 552 25 Sep2021 at23:59:23 pairs Density1 (no. of ties) 0.39 (28) 0.19 (39) 0.25 (189) 0.22 (113) 0.15 (257) 0.25 (138) Oryx Reachability2 100% 100% 79% reachable by 96% & 21% 100% 100% 100%

07 14,673 51(4), 2017, , reachable by # 18% Mean geodesic distance3 1.71 2.12 2.06 2.01 2.37 1.87 Network centralization index4 49% 69% 13% 43% 19% 49%

, subjectto theCambridgeCore termsofuse,available at Betweenness index – – – – – –

– Range 0 30 0 131 0 109 0 212 0 365 0 256 683 Mean 6 15 23 22 55 20 ©

06Fua&FoaItrainldoi:10.1017/S0030605316000454 International Flora & Fauna 2016 Sum 51 218 634 502 2319 480 % scored 0 33 53 39 52 26 46 No. of subgroups5 447697  Density equals the sum of the ties present divided by the number of possible ties.  The percentage of actors reachable by all others.  If the network is moderately dense the geodesic distances are generally small, suggesting that information may travel more rapidly in the network.  The higher the value the more centralized the network.  Groups of actors that are well-connected to each other, but with few connections between the groups. Australian bird conservation networks 679

TABLE 4 Potential influences (+, positive; −, negative) of the network attributes on the social networks for each of the six study species. Strength of attribute is indicated by intensity of shading.

Alligator Rivers yellow Capricorn Orange-bellied Swift Carnaby’s Baudin’s Attribute Metric chat yellow chat parrot parrot black-cockatoo black-cockatoo Collective mem- Density + −− −−− ory/experiences Reachability + + − ++ + Diversity No. of −−++++ subgroups Density −−++++ Redundancy Betweenness −−− −+ − Density + −− −−− Learning Betweenness − + −−+ − Centralization +/−− ++/− ++/− Reachability + + − ++ + Adaptive Centralization +/− + − +/−− +/− capacity Density − ++ +++ Reachability + + − ++ + Trust Betweenness − + −−−− Density + −− −−−

members than illustrated in Fig. , but those additions would The network had low density, with some actors not reach- be pendants (i.e. nodes connected to the network by a single able by others, high betweenness (linkage) scores of some tie). This is not to deny the potential significance of these actors and a low level of centralization. As a subgroup the nodes in occasionally undertaking supportive actions for recovery team was characterized by strong ties between management of the species, but the interview data estab- group members, but recovery team members recognized lished their peripheral role. others as part of a broader network with interests in conservation of the parrot. Alligator Rivers yellow chat Swift parrot The small network for this taxon had weak ties because of the infrequency of interactions between the actors, and the lack of All actors in this network were reachable by others but dens- coordination. There were no actions being implemented for ity was relatively low. The data suggested diversity was likely the recovery of the taxon at the time of the interviews. to be high, resulting from a degree of separation of subgroups supported by a low density of ties, with some actors having high betweenness scores. Centralization was neither Capricorn yellow chat high nor low. This network had weak ties to the timber in- This relatively small network was characterized by a low dustry despite the industry’s importance to conservation of density of ties and high centralization. Although four sub- the parrot in its breeding grounds. groups were identified, all were connected to one centralized actor. Only three individuals were not linked directly to that Carnaby’s black-cockatoo actor, and each was separated from the actor by only one node. Without a formal recovery team that meets regularly, Everyone in the network was reachable by others but there the network relied on funded projects to undertake conser- was a low density of ties, numerous subgroups and a low vation actions and maintain communication among net- level of centralization. The betweenness scores suggested work members. The lack of regular meetings and the high redundancy. There were many strong ties throughout reliance on informal and opportunistic communications the network, including beyond the recovery team. could be attributed to not having a formal recovery team, despite having a recovery plan. Baudin’s black-cockatoo

Orange-bellied parrot This network resembled the swift parrot network. All actors were reachable by others but density was relatively low and Diversity within this network was high, resulting from a de- there was a low density of ties with well-separated sub- gree of separation of groups and low overall density of ties. groups. Some actors had high betweenness scores but

FIG. 1 Management networks for six threatened Australian bird taxa: (a) Alligator Rivers yellow chat Epthianura crocea tunneyi, (b) Capricorn yellow chat Epthianura crocea macgregori, (c) orange-bellied parrot Neophema chrysogaster, (d) swift parrot Lathamus discolor, (e) Carnaby’s black-cockatoo Calyptorhynchus latirostris and (f) Baudin’s black-cockatoo Calyptorhynchus baudinii. Square nodes represent the individuals in the network; black nodes are recovery team members and grey nodes others. Lines represent ties between nodes, with thicker lines signifying stronger ties, based on individuals’ responses regarding the importance of others in the network.

centralization was neither high nor low. The network was responsibility for the subspecies’ conservation (Ainsworth, characterized bythe dominance ofstategovernment represen- ; Ainsworth et al., ). It may not be a coincidence tatives (%). Illegal killing of cockatoos by orchardists is con- that there appears to have been substantial, previously un- sidered to be a significant threat (Chapman, )but detected, decline in the last decade (Kyne & Jackson, ). although there had been ad hoc engagement with the fruit- All other taxa had a key individual or a recovery team growing industry in the past, maintaining communication that facilitated coordination. The Capricorn yellow chat had proved difficult and industry representation was lacking, had an individual champion who undertook actions, facili- thus limiting expectations of progress in threat mitigation. tated actions and/or advocated for actions to be implemented. However, high scores for centralization and density are Discussion not inherently conducive to good outcomes (Prell et al., ), as reliance on an individual champion is tenuous The coordination that is necessary for networks to contrib- and can be characterized by ad hoc actions, meetings and ute effectively to the conservation of a taxon was lacking exchanges of information. The network for the Capricorn for the Alligator Rivers yellow chat. A separate analysis yellow chat showed strong adaptive capacity, the only attri- of the values people hold for this subspecies found that, bute that scored positively among all measures for the two despite legislative obligations, no-one felt any individual networks that did not have recovery teams.

In cases where recovery teams had been established to can be done by recovery teams recognizing the significance undertake coordination of taxon conservation, subgroups of social networks, investing resources to improve under- constituted the basis of relevant networks. These subgroups standing of their existing networks and undertaking analysis were well-connected, had high densities of ties, and the ac- to develop favourable capabilities by building on their tors within them had higher betweenness scores, an attri- strengths and minimizing weaknesses. Such analysis could bute that generally contributes positively to the be undertaken both proactively, when initiating conservation effectiveness of networks (Prell et al., ). All the net- programmes for threatened species, and retrospectively for works with recovery teams had consistently high levels of di- programmes that are already established. The tools of analysis versity of actors, who probably contributed different are readily available, along with guidelines on how to inter- knowledge to the networks and thus increased the potential pret the results. As is evident from our findings, strengths for innovation (Bodin et al., ). All but one of the net- and weaknesses of network structures are often two sides of works with teams demonstrated strong adaptive capacity, the same coin, and the ideal network requires a balance tai- suggesting greater ability to manage new knowledge of the lored to the problem at hand rather than simply an increase in taxa more effectively or accommodate new information on resources. In fact network structures need to be optimized ac- threats and changing conditions. These three networks also cording to the resources that may be available. showed strong learning capacities, facilitating increased and Social network analysis can be a powerful managerial improved knowledge. tool for identifying issues that are hindering the effective- The presence of a recovery team within a network provided ness and capacity of a group, and to identify the specific be- focus, encouraged coordination of actions and facilitated the haviours and organizational elements that require flow of information within the network and beyond. Other modification if the group is to deliver conservation action benefits arose because the attributes of effective recovery effectively. An effective social network is likely to be particu- teams are useful for the functioning of networks (Holmes, larly important where species conservation is underfunded ); for example, the practical commitment for a recovery and relies heavily on volunteers. This study investigated the teamtomeetonaregularbasisincreasesthelikelihoodofcom- traits that can influence the efficiency and effectiveness of munication extending to others in the network. If a recovery social networks for six threatened bird taxa. We identified team has a broad representation of stakeholders, this can potential strengths and weaknesses in each of the networks also be beneficial for the diversity and learning of the network and demonstrated how attributes of the social networks as- (Bodin et al., ). Maintaining networks requires invest- sociated with each bird have the potential to influence out- ment of time and energy (Cross et al., a). The weaknesses comes for the conservation of threatened species. of networks with recovery teams may include the potential for Investment in social network analysis of recovery teams, a lack of trust and having little redundancy (replication of lin- or an understanding of social network theory when creating kages), which are characterized by having low betweenness such teams, offers the potential to increase their effective- scores and low density (Bodin et al., ). ness, especially if resources are constrained. Information, resources and perspectives tend to be more similar within, but different between, social networks. To improve the diversity of information and perspectives, Acknowledgements stakeholder representation and potential access to more re- We are indebted to all those working on the recovery of sources, recovery team membership could be strengthened threatened birds across Australia who participated voluntar- by becoming more representative of the various stakeholder ily in this study, and to Dr Gillian Ainsworth for collabor- clusters within the social networks (Reagans et al., ). ation in the field. We thank Claire Johnston for assistance Social network analysis has been used extensively for the with the figure. We obtained ethics approval for the study purposes of diagnosing and strengthening organizations from the Behavioural and Social Sciences Ethical Review (Cross et al., b; Scott, ). Each of the groups respon- Committee of the University of Queensland to undertake sible for, or with an interest in, conservation of the six threa- the study. Funding was provided by the Australian tened species examined here could benefit from Research Council as part of project LP, BirdLife understanding the vulnerabilities inherent in current net- Australia and Biosis. work structures and examining potential approaches to strengthening them. Recovery plans aim to facilitate recovery activities by providing a planned and logical framework Author contributions for action, including specific, measurable, achievable, relevant and time-bound objectives. Social networks could be TQH undertook the interviews, performed the primary ana- honed to improve communication, decision making, coord- lysis and drafted the initial manuscript, BWH, HPP and ination and implementation of conservation actions STG contributed to research design, and STG and BWH fi- (Vance-Borland & Holley, ; Mills et al., ). This nalized the article.

References Calyptorhynchus banksii naso) recovery plan. Western Australia Department of Environment and Conservation, Perth, Australia.  ADGER, W.N., BROWN,K.&TOMPKINS, E.L. () The political CHILVERS,J.&EVANS,J.( ) Understanding networks at the economy of cross-scale networks in resource co-management. science–policy interface. Geoforum, , –. Ecology and Society, , http://www.ecologyandsociety.org/vol/ COMMONWEALTH THREATENED SPECIES SCIENTIFIC COMMITTEE iss/art. () Approved Conservation Advice for Epthianura crocea tunneyi AINSWORTH, G.B. () Valuing birds: understanding the relationship (Yellow Chat (Alligator Rivers)). Http://www.environment.gov.au/ between social values and the conservation of Australian threatened cgi-bin/sprat/public/publicspecies.pl?taxon_id= [accessed  avifauna. PhD thesis. Charles Darwin University, Darwin, December ]. Australia. CRONA,B.&HUBACEK,K.() The right connections: how do AINSWORTH, G.B., ASLIN, H.J., WESTON, M.A. & GARNETT, S.T. social networks lubricate the machinery of natural resource () Do social values influence levels of conservation effort in governance? Ecology and Society, , http://www.ecologyandsociety. threatened species? The case of two Australian chats. Oryx, , – org/vol/iss/art/. . CROSS, R., BORGATTI, S.P. & PARKER,A.(a) Making invisible ALEXANDER, S.M., ARMITAGE,D.&CHARLES,A.() Social work visible: using social network analysis to support strategic networks and transitions to co-management in Jamaican marine collaboration. California Management Review, , –. reserves and small-scale fisheries. Global Environmental Change, , CROSS, R., NOHRIA,N.&PARKER,A.(b) Six myths about –. informal networks—and how to overcome them. Sloan  – ARMITAGE, D., DE LOË,R.&PLUMMER,R.() Environmental Management Review, , . governance and its implications for conservation practice. DEPARTMENT OF THE ENVIRONMENT () Environment Protection Conservation Letters, , –. and Biodiversity Conservation Act . Https://www.environment.   ARMSTRONG,M.() The Yellow Chat Epthianura crocea tunneyi in gov.au/epbc [accessed June ]. Kakadu National Park. Parks Australia (North), Darwin, Australia. DEPARTMENT OF ENVIRONMENT AND CONSERVATION () ’ BIRDLIFE INTERNATIONAL () Zanda baudinii. The IUCN Red List Carnaby s Cockatoo (Calyptorhynchus latirostris) Recovery Plan. of Threatened Species : e.TA. Http://dx.doi. Department of Environment and Conservation, Perth, Australia. org/./IUCN.UK.-.RLTS.TA.en DEPARTMENT OF SUSTAINABILITY,ENVIRONMENT,WATER, [accessed  June ]. POPULATION AND COMMUNITIES () Recovery plans. Http:// BIRDLIFE INTERNATIONAL () Zanda latirostris. The IUCN Red www.environment.gov.au/biodiversity/threatened/recovery.html List of Threatened Species : e.TA. Http://dx. [accessed  December ]. doi.org/./IUCN.UK.-.RLTS.TA.en FOWLER, J.H. & CHRISTAKIS, N.A. () Dynamic spread of [accessed  June ]. happiness in a large social network: longitudinal analysis over   BIRDLIFE INTERNATIONAL (a) Lathamus discolor. The IUCN Red years in the Framingham Heart Study. British Medical Journal, , List of Threatened Species : e.TA. http://dx.doi. a. org/./IUCN.UK.-.RLTS.TA.en GARNETT, S., SZABO,J.&DUTSON,G.() Action Plan for [accessed  June ]. Australian Birds . CSIRO, Melbourne, Australia.  BIRDLIFE INTERNATIONAL (b) Neophema chrysogaster. The IUCN HIGGINS, P. (ed.) ( ) Handbook of Australian, New Zealand and Red List of Threatened Species : e.TA. Http:// Antarctic Birds. Volume . Parrots to Dollarbird, Oxford University dx.doi.org/./IUCN.UK.-.RLTS.TA.en Press, Melbourne, Australia. [accessed  June ]. HOLLAND,J.() Tools for Institutional, Political, and Social BODIN,Ö.&CRONA, B.I. () The role of social networks in natural Analysis of Policy Reform: A Sourcebook for Development resource governance: what relational patterns make a difference? Practitioners. The World Bank, Washington, DC, USA. Global Environmental Change, , –. HOLMES, T.Q. () Analysis of the institutional arrangements for the ’ BODIN, Ö., CRONA,B.&ERNSTSON,H.() Social networks in management of Australia s threatened birds. PhD thesis. University natural resource management: what is there to learn from a of Queensland, St Lucia, Australia. structural perspective? Ecology and Society, , http://www. HOUSTON,W.&MELZER,A.() Yellow Chat (Capricorn ecologyandsociety.org/vol/iss/resp/. subspecies) Epthianura crocea macgregori Recovery Plan. Report to BORGATTI, S.P., CARLEY, K.M. & KRACKHARDT,D.() On the Department of the Environment, Water, Heritage and the Arts, robustness of centrality measures under conditions of imperfect Canberra. Queensland Environmental Protection Agency, Brisbane, data. Social Networks, , –. Australia.  BORGATTI, S.P., EVERETT, M.G. & FREEMAN, L.C. () Ucinet  for JEPSON,P.( ) Saving a species threatened by trade: a network study Windows: Software for Social Network Analysis. Analytic of Bali starling Leucopsar rothschildi conservation. Oryx, , –. Technologies, Lexington, USA. JEPSON, P., MARUA,M.&BUCKINGHAM,M.() What is a  – BOTTRILL, M.C., WALSH, J.C., WATSON, J.E.M., JOSEPH, L.N., conservation actor? Conservation and Society, , .  ORTEGA-ARGUETA,A.&POSSINGHAM, H.P. () Does recovery KEAST,A.( ) The relationship between seasonal movements and planning improve the status of threatened species? Biological the development of geographic variation in the Australian Chats, Conservation, , –. (Epthianura Gould and Ashbyia North (Passeres: Muscicapidae,  – BUTTS, C.T. () Revisiting the foundations of network analysis. Malurinae)). Australian Journal of Zoology, , . Science, , –. KEAST, R., MANDELL, M.P. & AGRANOFF, R. eds () Network CAMPBELL, N.A. & SAUNDERS, D.A. () Morphological variation in Theory in the Public Sector: Building New Theoretical Frameworks. the white-tailed black cockatoo, Calyptorhynchus baudinii,in Routledge, London, UK. Western Australia: a multivariate approach. Australian Journal of KOLIBA, C.J., MILLS, R.M. & ZIA,A.() Accountability in Zoology, , –. governance networks: an assessment of public, private, and CHAPMAN,T.() Forest black cockatoo (Baudin’s cockatoo nonprofit emergency management practices following Hurricane Calyptorhynchus baudinii and forest red-tailed black-cockatoo Katrina. Public Administration Review, , –.

KOSSINETS,G.() Effects of missing data in social networks. Social SCHODDE,R.&MASON, I.J. () The Directory of Australian Birds: Networks, , –. Passerines. CSIRO, Melbourne, Australia. KYNE, P.M. & JACKSON, M.V. () Surveys for Yellow Chat Epthianura SCOTT,J.() Social network analysis: developments, advances, and crocea tunneyi on the South Alligator River Floodplain, Kakadu prospects. SOCNET, , –. National Park, . A report to Parks Australia, Darwin, Australia. STOJANOVIC, D., WEBB, M.H., ALDERMAN, R., PORFIRIO, L.L. & LAUBER, T.B., DECKER, D.J. & KNUTH, B.A. () Social networks HEINSOHN,R.() Discovery of a novel predator reveals extreme and community-based natural resource management. but highly variable mortality for an endangered migratory bird. Environmental Management, , –. Diversity and Distributions, , –. LUKASIEWICZ, A., DAVIDSON, P., SYME, G.J. & BOWMER, K.H. () STOKER,G.() Governance as theory: five propositions. How the social construction of the environment affects people’s International Social Science Journal, , –. reactions to water policy. Australasian Journal of Environmental SWIFT PARROT RECOVERY TEAM () National Swift Parrot Management, , –. Recovery Plan –. Department of Primary Industries, Water MILLS, M., ÁLVAREZ-ROMERO, J.G., VANCE-BORLAND, K., COHEN, and Environment, Hobart, Australia. P., P RESSEY, R.L., GUERRERO, A.M. & ERNSTSON,H.() Linking TOMPKINS, E.L. & ADGER, W.N. () Does adaptive management of regional planning and local action: towards using social network natural resources enhance resilience to climate change? Ecology and analysis in systematic conservation planning. Biological Society, , http://www.ecologyandsociety.org/vol/iss/art. Conservation, , –. VANCE-BORLAND,K.&HOLLEY,J.() Conservation stakeholder ORANGE-BELLIED PARROT RECOVERY TEAM () National network mapping, analysis, and weaving. Conservation Letters, , Recovery Plan for the Orange-Bellied Parrot. Department of Primary –. Industries and Water, Hobart, Australia. WASSERMAN,S.&FAUST,K.() Social Network Analysis: OSTROM,E.() Understanding Institutional Diversity. Princeton Methods and Applications. Cambridge University Press, Cambridge, University Press, Princeton, USA. UK. OTTE,E.&ROUSSEAU,R.() Social network analysis: a powerful WEY, T., BLUMSTEIN, D.T., SHEN,W.&JORDÁN,F.() Social strategy, also for the information sciences. Journal of Information network analysis of animal behaviour: a promising tool for the study Science, , –. of sociality. Animal Behaviour, , –. PRELL,C.() Social Network Analysis: History, Theory & WOODWARD,E.() Social networking for Aboriginal land Methodology. SAGE Publications Ltd, London, UK. management in remote northern Australia. Australasian Journal of PRELL, C., HUBACEK,K.&REED,M.() Stakeholder analysis and Environmental Management, , –. social network analysis in natural resource management. Society & Natural Resources, , –. REAGANS, R., ZUCKERMAN,E.&MCEVILY,B.() How to make Biographical sketches the team: social networks vs. demography as criteria for designing effective teams. Administrative Science Quarterly, , –. T IM H OLMES is a former senior public servant specializing in conser- REED, M.S., GRAVES, A., DANDY, N., POSTHUMUS, H., HUBACEK, K., vation management, with a particular focus on the management of MORRIS, J. et al. () Who’s in and why? A typology of threatened bird species. B RIAN H EAD’s recent work includes projects stakeholder analysis methods for natural resource management. on the contribution of social science to complex problem-solving, and Journal of Environmental Management, , –. the role of rigorous evidence in policymaking. H UGH P OSSINGHAM SAUNDERS, D.A. () The availability of tree hollows for use as nest directs the Australian Research Council Centre of Excellence for sites by white-tailed black cockatoos. Wildlife Research, , –. Environmental Decisions and the National Environmental Science SAUNDERS, D.L. & TZAROS, C.L. () National Recovery Plan for Swift Programme Threatened Species Recovery Hub. S TEPHEN G ARNETT Parrot Lathamus discolor. Birds Australia, Melbourne, Australia. has worked on Australian threatened birds for the last  years.