Stakeholder discourse coalitions and polarisation in the hen harrier conservation debate in news media

1. Conservation conflicts are complex and can be deep-rooted, with stakeholders holding entrenched policy positions. The actors involved producing verbal inter-connected interactions that form policy debates. Thus, conservation debates can be viewed as network phenomena with stakeholders forming coalitions in support of, or opposition to, certain policies and practices. 2. We used Discourse Network Analysis of print media to investigate the structure and dynamics of the stakeholder debate around the management of hen harriers Circus cyaneus , a bird of prey at the centre of a long-standing conservation conflict in the United Kingdom. 3. We aimed to determine whether the structure of discourse coalitions changed among the diverse aspects of the debate and whether the polarisation of the debate has changed through time. Our search and selection criteria led to the analysis of 737 statements within 131 newspaper articles published from August 1993 to December 2019. 4. We show that, while the discourse network of the overall debate is quite unstructured


| INTRODUC TI ON
Conservation conflicts are increasing in numbers and intensity (Redpath et al., 2015). These are disagreements between parties over conservation objectives in the context of perceived imbalance of stakeholder interests (Redpath et al., 2013). Conservation conflicts lie within unique and often complex socio-ecological contexts (Young et al., 2010) and are shaped by human-wildlife and human-human interactions (Redpath et al., 2013). Therefore, conservation scientists have explored drivers of conflict, mitigation measures and conflict resolution processes at different scales and in different contexts. Studies have focused on, for example, drivers of human tolerance towards large carnivores such as wolves Canis lupus and brown bears Ursus arctos in Italy (Marino et al., 2020), alert systems to reduce livestock losses to lion Panthera leo predation in Botswana (Weise et al., 2019), and conflict transformation in stakeholder conflicts about mountain lions Puma concolor in the United States and elephants in the African continent (Madden & McQuinn, 2014).
Conservation conflicts can influence global conservation policies, both as drivers of policy development and as challenges that can cause biodiversity loss (Carmen et al., 2015). Underpinning many conflicts are differences in conceptions of nature, such as its state and importance, which can contribute to differing policy narratives and discourses (Adams, 2015). These conflicts can also emerge as struggles for power where stakeholders attempt to establish the dominance of a certain discourse in policy arenas (Buchanan, 2013). Those lacking the power to influence formal decision-making processes may turn instead to other means (e.g. social media, protest), which could result in conflict escalation (Crowley et al., 2017). Stakeholders may seek to empower their positions across multiple media. This is done to support their respective political and policy interests by making strategic use of their discourses and diverse types of knowledge, for example, research-based and technical evidence, local and anecdotal, often to the detriment of opposing parties in decision-making processes (Buchanan, 2013;Hodgson et al., , 2019. As a result, conservation conflicts can persist (Redpath et al., 2013) and evolve through escalation and de-escalation over time (Crowley et al., 2017). Media can also play an active role in the exacerbation of conflicts as, for example, journalists might seek to construct stories presenting opposition between sides (Fiorina et al., 2005).
More recently, social media have increasingly played a similar role by facilitating political polarisation (Van Bavel et al., 2021). For these reasons, it is pivotal to investigate conservation conflicts in media arenas and in relation to their temporal scale.
In politics, actors such as legislators and interest groups produce verbal interactions about policies that form policy debates or political discourses (Leifeld, 2017). These interactions include public statements in support of, or opposition to, a given policy.
Discourses are interdependent, potentially weighted and directional and, hence, can be characterised as networks (Leifeld, 2017).
Investigating these networks allows us to consider some of the mechanisms shaping policy debates in a joint and systematic way, such as coalition formation and framing (Leifeld, 2017). From a network perspective, these mechanisms address respectively the clustering of actors and content. Frames, in particular, are the repertoires of categories and concepts through which actors interpret phenomena (Buijs et al., 2011).
In policy debates, these frames consist of clusters of policy preferences, beliefs, or justifications, and can be associated with impasse and communication between actors (Keenan et al., 2020).
Actors holding similar belief systems, reflected by preferences for policy instruments and lines of evidence, form coalitions that compete to achieve different policy designs (Sabatier, 1988). However, it is through their discourse that actors give meaning to phenomena and form discourse coalitions, where these are 'the ensemble of a set of story lines, the actors that utter these story lines, and the practices that conform to these story lines, all organized around a discourse' (Hajer, 1993(Hajer, , 1995. Discourse coalitions are thus the result of actors positioning through their statements, and function to provide arguments and narratives in support of, or opposition to, certain policies. To gain insight into relational aspects of conservation issues, conservationists have used social network analysis (SNA), which provides a set of tools for the study of social structures and their relational characteristics (Scott, 2012). For example, SNA allowed the study of relations and communication between conservation stakeholders protecting a biota shared across different countries (Moshier et al., 2019) and combatting illegal wildlife trade (Gogaladze et al., 2020), and has been proposed to inform decision-making in systematic conservation planning (Mills et al., 2014). In relation to conservation conflicts, SNA has been used to disclose the high degree of polarisation between organisations in the conflict over hunting migratory birds in Malta (Veríssimo & Campbell, 2015). However, to date, we have only a limited understanding of discourse as a network phenomenon in conservation conflicts and impasses related to management and policy-making.
The decade-long debate over the conservation and management of the hen harrier Circus cyaneus in the United Kingdom (Thirgood & Redpath, 2008) is a useful example to understand the role of actors and their discourses in shaping conservation conflicts over time. High-profile actors have likely perpetuated the conflict through their discourses in the media, possibly hindering mitigation processes and exacerbating polarisation (Hodgson et al., 2019). Hodgson et al. ( , 2019 investigated elements of the debate in internet media using discourse analysis. These studies found divergent interpretations and use of research-based knowledge and differing storylines portraying other stakeholders and their role in illegal killing. These aspects, in turn, are likely the result of strategic efforts on the part of raptor conservation and shooting organisations to influence policy-making processes concerning land use and licensing of grouse shooting or raptor control. Recognising that discourse is crucial in driving, shaping and potentially alleviating the hen harrier conservation conflict, our study aimed to investigate the structure and dynamics of stakeholder debate on hen harrier conservation and management. To this end, we used discourse network analysis (DNA), a methodology developed 'to describe the structure of political discourses and infer their generative processes' (Leifeld, 2017). DNA achieves this by addressing the configuration of actors along with the structure of the concepts in policy debates (e.g. policy beliefs, preferences, justifications) through a combination of content analysis and SNA. DNA operationalises the discourse coalitions of proponents and opponents of policy measures and highlights their structure and complexity. Most importantly, DNA allows the longitudinal and systematic study of policy debates and their mechanisms. This can reveal dynamics such as polarisation through time and also promote theory-building processes (Fisher & Leifeld, 2019;Leifeld, 2017). Therefore, we chose this methodology to provide valuable insight into enduring conservation conflicts such as the hen harrier case study.
DNA studies have explored a diverse range of policy debates in different domains, topics and arenas including debates in national newspapers regarding the German pension system (Leifeld, 2013) and, in the United Kingdom, the Soft Drinks Industry Levy (Buckton et al., 2019) and the Minimum Unit Price on alcohol Hilton et al., 2020). In the environmental domain, DNA studies have largely investigated climate change discourses in arenas such as sessions of the US Congress (Fisher et al., 2013;Fisher & Leifeld, 2019), verbatim reports of proceedings from the Italian parliament (Ghinoi & Steiner, 2020), testimonies on law proposals in Finland , and stakeholder statements in Finnish and Canadian newspapers . The potential of this methodology has yet to be realised in instances of conservation conflicts.
The aim of our study was to delve into the hen harrier debate due to the complex, entrenched nature of the conflict and its associated discourse. In particular, we sought to dissect the debate into its conceptual categories and study it through a longitudinal perspective as we expected to disclose dynamics that a coarser discourse analysis would not reveal. We did this by using the DNA methodology to analyse the coverage of the hen harrier conservation and management in the news media, one of the arenas of the debate. We tested two predictions: first, that discourse network clusters (or coalitions) and their dynamics vary with conceptual categories of discourse that structure the debate; second, that the level of polarisation between discourse coalitions has increased over time, where polarisation is the tendency of clusters to separate and not overlap in regard to key concepts.
Our application of the DNA focusses on revealing debate dynamics and attributes, such as highly polarised conceptual categories that could then be associated with the persistence of the conflict.
This insight could inform both the design of mitigation and resolution strategies that target conflicting organisational discourses and, ultimately, benefit conservation programmes.

| RE S E ARCH CONTE X T: THE HEN HARRIER CONS ERVATION CONFLIC T IN THE UK
The hen harrier conservation conflict stems fundamentally from depredation by harriers of game birds, primarily red grouse Lagopus l. scoticus, and the illegal killing of harriers by people seeking to increase grouse populations for sport shooting (Etheridge et al., 1997).
More specifically, those advocating for harrier protection have denounced grouse moor managers for the persistent illegal killing of harriers and have been reluctant to implement active forms of harrier management, while grouse moor managers have opposed stronger forms of legal protection for harriers (Thirgood & Redpath, 2008;Young et al., 2010). Evidence from the natural sciences, for example on the illegal killing of raptors, the impact of predation upon grouse populations, and the sustainability of current approaches to grouse shooting, has proven insufficient to identify or gather support for solutions given the entrenched positions of the main stakeholders (Thirgood & Redpath, 2008). Despite a long history of conservation measures (Thirgood & Redpath, 2008) and the recent decline of territorial pairs of harriers in Great Britain (Wotton et al., 2018), the hen harrier conflict appears to have worsened and become more polarised both in England and in Scotland St John et al., 2019). This includes actions such as diversionary feeding, a brood management scheme on moorland, and a reintroduction in Southern England facilitated by the statutory nature conservation body Natural England.
Diversionary feeding is the provision of an alternative food source to divert harriers' predation on game birds. Brood management consists of the translocation of chicks hatched on participating game estates to unoccupied habitats once a certain density of hen harriers is reached (density based on Elston et al., 2014). The reintroduction seeks to translocate birds from continental Europe to suitable habitats in Southern England to create a viable and self-sustaining population, which is then hoped to expand its range. However, support for these actions varies among and within stakeholder groups (St John et al., 2019). Some actors from the field sports community support brood management and the reintroduction. Some of the conservation communities focusing on the protection of birds oppose the brood management scheme and do not support the reintroduction. Similarly, brood management was also opposed by those protecting raptors specifically, while the reintroduction by those protecting birds apart from raptors (St John et al., 2019).

| Data sourcing
The first step to test our predictions and understand the debate in the news media was to conduct a qualitative content analysis of stakeholders' statements reported in news articles. Among print media, we considered newspapers as the most comprehensive form to investigate stakeholder discourse. Therefore, we sourced newspaper articles published in the United Kingdom from the online Nexis database (LexisNexis, 2021). The database provides news from several sources including UK national and regional newspapers as well as international newspapers and newswires. After a tentative search, the search terms used were 'hen harrier' W/10 (within 10 words) 'management', 'hen harrier' AND 'translocation' and 'hen harrier' AND 'reintroduction'.
The first term, 'hen harrier' W/10 (within 10 words) 'management', was selected to reduce the number of newspaper articles mentioning the two terms but in different contexts, unrelated to the management of the species. As a result, these terms allowed coverage of the broader debate on the management of the species while focusing on the most divisive aspects of the HHAP, that is, the brood manage-

| Data coding
We exported the selected articles to the software Discourse Network Analyser (2.0 beta 25) (Leifeld, 2019). By default, the software allows coding statements for four variables: the person making a statement, the organisation to which the person is affiliated, the concept representing statement content and agreement over the concept (binary variable: yes/no). Additionally, we coded the year of publication and geographical region referred to in the statements, that is, England or Scotland. When the region was not explicit, the first author (FM) inferred it from the context of the article. Inductive qualitative-content coding led to the development of a coding framework for concept structured into categories and sub-categories (Leifeld, 2017). Two researchers (FM and NAWF) independently double-coded a random 10% sample of newspaper articles and then discussed similarities and inconsistencies in the codes used. This step allowed to increase validity by reducing personal bias and the risk of missing concept and, ultimately, led to the development of a coding framework. The main author then conducted the coding of the news articles. During the process, the concepts and their categorisation were reviewed with the support of a third author (SLC). As a result, 737 statements were coded and three main conceptual categories were identified. We report the most common concepts per conceptual category in Table 1. Additionally, we report the coding framework in Table S1 (Appendix S1), a full list of the actors involved in the debate in Table S2 (Appendix S2), and a full breakdown of the statements coded in Table S5 (Appendix S2).

| Prediction testing, data analysis and visualisation
We imported the coded and structured data to the R software (R Core Team, 2019) for network development and analysis using the package rDNA (Leifeld & Henrichsen, 2019). We removed duplicated statements in the same article (e.g. an actor expressing repeated agreement/disagreement over the same concept) as the number of times a statement occurs in a newspaper could depend on a journal or journalist's agenda or reporting style rather than the actor's activity.
Discourse networks were based on weighted matrices with debate actors arranged in rows and columns and the matrix entries corresponding to the level of agreement or disagreement on the coded concepts.
The matrices were then visualised as networks: the nodes representing the actors and the weighted edges, that is, the links, representing the overall agreement between pairs of actors. Specifically, we applied the subtract network method (Leifeld, 2017) which measures stakeholder conceptual similarity. This involves the subtraction of a conflict matrix, where edge weight reflects the extent to which actors have opposing agreement patterns, from a congruence matrix, where edge weight reflects the extent to which actors have similar agreement patterns (either co-support or co-rejection of concepts).
We produced networks for the whole debate and for each of the conceptual categories identified through the content analysis to test our prediction that coalitions would depend on structural categories within the wider hen harrier debate. In each subtract network, we normalised actors' activity through average activity normalisation (Leifeld, 2017). Average activity normalisation involved dividing edge weights by the average number of concepts stated by any pair of actors. Thus, we ensured that edge weights reflected only argumentative similarity while accounting for the statement rate of vocal actors, which are more likely to agree or disagree with other actors. To identify stakeholder coalitions, we analysed the modularity (Newman & Girvan, 2004) of the networks. Modularity is a measure of community structure representing the tendency of network nodes to group in different clusters (Newman & Girvan, 2004). By assessing modularity, we investigated the tendency of the network actors to fall into distinct discourse coalitions. Specifically, modularity measures the fraction of within-community edges minus the expected value of the fraction within the same community if edges were random. Modularity ranges between −0.5 and 1.0, where high modularity >0.3. Positive values suggest the existence of community structure (e.g. discourse coalitions), namely groups of closely connected nodes or vertices (e.g. stakeholders) with limited connection to other groups (Newman, 2006;Newman & Girvan, 2004). We did this using the software Visone (Visone Project Team, 2018) to which we imported networks also for visualisation. We used one-mode subtract networks, namely networks with a single type of node (i.e. actors), and retained only positive edge weights to focus on argumentative similarity. This was done to remove the bias for network visualisation due to the presence of negative ties (Leifeld, 2017). We used the Louvain community detection algorithm to highlight stakeholder coalitions in network diagrams.
We tested our second prediction on polarisation by analysing its levels and variation over time. The less the coalitions are bridged by intermediate actors, the higher the polarisation. To assess polarisation we relied on modularity. We used the function 'dna_multiclust' from the rDNA package. This maximises modularity by applying 14 clustering techniques to measure modularity and retaining the highest value of network modularity obtained. We did not pre-determine a specific

| Ethical statement
The project received ethical approval from the University of Exeter

| Conceptual structure of the debate
The debate over the conservation of the hen harrier and its management was based upon three main conceptual categories: Problems, Solutions, and Reactions (Table 1 and Table S1). Concepts associated TA B L E 1 Top five most frequent concepts in each conceptual category (Problems, Solutions and Reactions) in the debate over the management and conservation of the hen harrier in England and Scotland based on news articles published from August 1993 to December 2019. Frequent instances of both agreement and disagreement indicate the level of polarisation for each concept. * HH = Hen harrier

| Actors and concepts composition
Over the years, a wide range of actors has participated in the conservation debate (Table S2) (Table 1). Problems and Solutions were the most divisive categories and exhibited higher levels of disagreement over their concepts (Table 1) Table S2).

| Discourse networks of the debate and differences
Network structures differed, and were more defined, in the discourse sub-networks rather than in the overall discourse network.
Actors clustered in more defined coalitions, shown by Louvain modularity values, and clustered differently depending on the conceptual category (Figure 3).
In the overall debate, characterised over the whole period and both countries, the discourse network was not clearly polarised.  Table S2 for the list of types of actors and full names.

| Recent developments
Our further investigation of discourse networks during the period 2014-2019 showed that the modularity of the discourse networks was on average higher in recent years than in the entire study period

| DISCUSS ION
Understanding stakeholder discourses related to conservation policy and practice is crucial to navigate towards conflict mitigation and effective conservation actions. We investigated stakeholder discourses surrounding the conservation and management of hen harriers in England and Scotland with a DNA of the newspaper media. We confirmed our predictions that discourse networks and their coalitions differed depending on the conceptual categories that constitute the debate. Specifically, the discourse was more polarised into coalitions when addressing the conservation problem and its solutions, rather than when delivering affective reactions, which many actors shared. We highlighted that coalitions have become more pronounced and the discourse more  England. Here, we discuss these dynamics and their potential drivers as well as other internal and external factors (e.g. stakeholder actions, events and debate arenas) that contribute to shaping conservation policy debates.

| The debate
A multitude of actors has contributed to the hen harrier debate.
However, only a subset was regularly engaged in the print media  (Hodgson et al., , 2019. Our study supports previous research, suggesting that focal organisations such as RSPB can institutionalise their views, meaning these become rooted in institutions as practices and ways of thinking (Hajer, 1993), and ultimately influence policy-making . Moreover, it hints at a shared effort from the game and countryside industry to counteract such a drive as shown by the discourse coalition that pro-shooting actors formed, and which remained almost unchanged, in all Problems and Solutions sub-networks.
The debate was also characterised by a geographical shift during the study period with an increase of statements referring to the conservation and management of hen harriers in England rather than Scotland. This could be the result of the persisting critical status of the hen harrier in England and, consequently, a shift in the focus of discourses. Despite an overall decline, most hen harriers in the United Kingdom (76%-80%) have been regularly found in Scotland (Hayhow et al., 2013;Sim et al., 2007;Wotton et al., 2018). In England, instead, the species almost disappeared with the number of territorial pairs decreasing from 19 in 1998 to 4 in 2016 (Hayhow et al., 2013;Sim et al., 2007;Wotton et al., 2018).

| Discourse network dynamics
Supporting our first prediction, discourse networks based on conceptual categories differed in terms of both their polarisation and the composition of coalitions. The divisiveness associated with defining the conservation problem and its solutions did not appear in the overall debate. Instead, it became evident only when these two categories were isolated from the discursive similarity of the affective reactions.
This might suggest the presence of a debate dynamic where part of discourse concepts counterbalance the divisiveness of other concepts. Moreover, some of the most relevant and active stakeholders On the left are the supporters, and on the right are the opponents of the brood management Solution, which is one of six actions of the Hen Harrier Action Plan. Node sizes reflect the number of times actors referred to the concept; node colours represent the types of actors, and edges show agreement between actors on the concept (see Table S2 for a list of types of actors and full names). . However, this could be the result of the gen- such as GWCT and SNH attempting to be 'middle-ground' groups . These actors were invisible in internet media as they did not contribute to narratives over specific themes (i.e. actions and statements of other actors, and illegal killing) in their organisational websites . Instead, in our study of print media they were visible but appeared in various coalitions  Table S2 for a list of types of actors and full names).
depending on the conceptual category. Second, actors might 'reinvent' themselves. In the policy debate over renewable energies in the United Kingdom, reinvention is a strategy adopted to reframe 'an old or polluting technology as new or innovative' (Johnstone et al., 2017). Actors associated with the natural gas industry actively reframed a non-renewable gas resource as a low-carbon and sustainable energy option. In this context, reinvention might have occurred across conceptual categories. For example, within Problems, some game and countryside organisations frequently claimed that the benefits associated with grouse shooting were not only socioeconomic but also ecological. This was a common concept in the debate and might have served to balance against the concept 'Game is associated with illegal killing', which was the most frequent Problems concept. The same reinvention could explain why a large majority of actors grouped in a single cluster in the affective discourse of the Reaction category. Through this strategy, actors might try to reduce or avoid conflict with other organisations, meanwhile engaging the public and policy-makers to shape their opinions thus supporting certain directions in policymaking.

| Polarisation
Confirming as it can reflect adherence to different policy paradigms without antagonism (i.e. segregation; Leifeld, 2020). However, given the acknowledged entrenched nature of the hen harrier conflict (Thirgood & Redpath, 2008), we believe that increased polarisation reflects increased antagonism. To some extent, the polarisation trend reflects the conflict curve common to social (and conservation) conflicts, where conflict can escalate due to the growth of iterative claiming and counter-claiming of those entangled in the debate (Crowley et al., 2017). Our result reinforces the suggestion that this stakeholder conflict was schismogenetic , where schismogenesis (sensu Brox, 2000 based on Bateson, 1935) is the process by which conflict is escalated through expressive competition that produces a recursive and escalating confrontation. This can originate when people sharing values and seeking different goals recur to political confrontation (Brox, 2000) but also when a lack of institutional protection of basic human rights characterises political negotiation (Harrison & Loring, 2014).
The polarisation of the hen harrier conservation debate is rooted, at least in part, in divergent values and value systems of the major actors involved (Hodgson et al., , 2019St John et al., 2019).
Wildlife value orientations range from mutualism (individuals viewing wildlife as deserving of rights and welfare similarly to humans) to utilitarianism/domination (individuals prioritising human mastery over wildlife and human well-being; Kellert, 1984;Manfredo et al., 2009). A substantial divergence of these orientations could be  (Cole, 2015;Ostrom, 2010).
Still, it is important to acknowledge that in certain policy debates, polycentricity can be used to produce policy blockage (Fisher & Leifeld, 2019 entirely. Overall, we argue that our analysis of the debate in its concepts can help design reconciliation or mitigation processes that target conceptual common ground and address criticalities such as divergent discourse elements and the roles of actors in the debates.   or in the US Congress (Fisher et al., 2013;Fisher & Leifeld, 2019). Internal factors concerning the actors themselves and their discursive contribution to the debate could explain this. Academics might intentionally focus on providing research results that could ease the conservation conflict rather than actively engaging in a confrontational debate. This might be the case given that the mitigation of the hen harrier conflict has been historically slow despite some substantial scientific (Thirgood & Redpath, 2008).

| Factors shaping conservation debates
Yet, this might hint at the recognised disconnect between academic and public debate (Lester & Foxwell-Norton, 2020). It is also important to consider that actors' roles can vary across arenas. Coalitions almost absent in a news media arena might be more prominent in non-media ones . For example, actors might intentionally go 'invisible' in the media, especially with entrenched divisions (Lester & Hutchins, 2012), opting instead to influence policy-makers through direct contact, that is, inside lobbying (Vesa et al., 2020).
External factors, originating from other actors, policy-making processes and policy arenas can also influence the role of actors and the debate itself. In the debate over climate change in the US congress, the failure of the cap-and-trade bill legislation (i.e. marketbased approach to control emissions) likely led to a vacuum of policy mechanisms, a lower involvement of congress members and, ultimately, an increased demand for scientist participation (Fisher et al., 2013;Fisher & Leifeld, 2019). In our study, the absence of a similar vacuum could have influenced the role of academics since other actors addressed a wide range of aspects of the conservation issue and potential solutions (e.g. HHAP).
Lastly, debate arenas are not independent and other arenas have likely contributed to shape the debate in news media and its polarisation. Social media, for example, facilitate the spreading of sensationalistic content from news media covering species such as large carnivores  and spiders , with the potential of affecting people's emotions, perceptions of risks, and attitudes towards these species. Similarly, in our study, social media platforms could have influenced both the actors involved and the journalists, thus reiterating the production and selection of conflictual discourses and concepts.

| Limitations and future directions
Our study contributes to a better understanding of conservation policy debates and conflicts. Nevertheless, it comes with limitations due to the scope of the analysis and the media source used. First, our DNA covered only part of the debate over hen harrier conservation by focusing on one of its arenas, that is, newspaper media. In other arenas such as social media platforms, discourse networks and their polarisation could be different as a result of different actors, concepts and agreement . Second, the targeted newspaper arena could be intrinsically biased as news media might offer only a skewed representation of conservation issues by favouring conflictual themes (Hughes et al., 2020;Niemiec et al., 2020). This is consistent with the known trends in news media of journalists relying on powerful sources (official dominance) and news being increasingly 'negative, dramatized, fragmentized, and personalized' (information biases) (Korthagen, 2013). Therefore, it is crucial to distinguish between real and media-exaggerated conflicts and coalitions.
Future discourse network analyses represent an opportunity not only to tackle these limitations and improve our understanding of conservation conflicts and their dynamics but also to expand understanding and application of DNA (Leifeld, 2020). Comparative studies of discourse networks across different arenas could contribute to a better understanding of the role of actors and their discourses, the biases of arenas and, hence, of the hen harrier debate as a whole. For example, it would important to investigate the actual power of dynamic coalitions and their influence on decisionmaking. The reason is that power and influence can vary in strength across different arenas ) and actors can shape public behaviour towards policy based on their position of power (Rinscheid, 2020). Future research should also broaden its scope by targeting debates over the management of different species, hence across similar policy domains, and incorporating inferential analysis (Brandenberger, 2019; Leifeld & Brandenberger, 2019).
Developments in this area should lead to an integrated study of value systems and DNA that could help understand the generative processes underlying discourse network dynamics in conservation conflicts. Systematic implementations of DNA could also provide an important monitoring tool. Research in the conservation domain can thus contribute to the need for prediction and systematic comparisons of discourse networks, which relies on a better understanding of mechanisms behind policy debates as dynamic networks (Leifeld, 2020).

ACK N O WLE D G E M ENTS
This study was supported by the University of Exeter and Natural England.

CO N FLI C T O F I NTE R E S T
None of the authors is affiliated or involved with any of the actors, either individuals or organisations, with any interest in the policy debate discussed in this study. However, the study has received financial support from Natural England as stated in the acknowledgements. Natural England has supported but not influenced the outcomes of this research. Sarah L. Crowley is an Associate Editor for People and Nature but was not involved in the peer review and decision-making process.

DATA AVA I L A B I L I T Y S TAT E M E N T
All the relevant data are available in the Nexis UK news online database.