Introduction
Conflict between carnivores and livestock farmers threatens predator populations and farmers’ livelihoods (Loveridge et al., Reference Loveridge, Wang, Frank, Seidensticker, Macdonald and Loveridge2010; Rust & Marker, Reference Rust and Marker2014). This conflict is difficult to resolve, partly because of complex social disagreements on governance options and goals (Clark et al., Reference Clark, Rutherford, Mattson, Clark and Rutherford2014). To reduce the problem effectively requires focusing on mitigating conflict not only between people and predators but also between various groups of people (Redpath et al., Reference Redpath, Bhatia and Young2015).
Historically, conservation of threatened species such as carnivores involved little participation from the local communities that were affected by wildlife management decisions (Brockington, Reference Brockington2002). This lack of participation can create tensions between stakeholders regarding how to mitigate the situation effectively (Thirgood & Redpath, Reference Thirgood and Redpath2008), which can sometimes result in communities revolting against management decisions (Goldman et al., Reference Goldman, Roque de Pinho and Perry2013). It is seen as increasingly important for wildlife managers to cooperate with the communities that share land with wildlife, and to involve them in management decisions. Community engagement has the potential to foster more amicable relations between all parties, which can lead to more socially accepted and sustainable management plans (Kittinger et al., Reference Kittinger, Bambico, Watson and Glazier2012).
This study focuses on human–predator conflict in Namibia, where carnivore populations have been increasing in recent decades, causing more frequent livestock depredation (NACSO, 2013). Official management of human–wildlife conflict is overseen by the government but the policy focuses almost exclusively on communal farmers who live on government land, and largely ignores freehold commercial farmers (Government of Namibia, 2009). Because of this skewed governance it is legal for commercial farmers to kill carnivores on their land if deemed a threat to human lives or property (Government of Namibia, 1975). This conflict requires immediate attention to ensure minimal damage to farmers’ livelihoods and recently restored carnivore populations. Creating a participatory management plan could help to solve this problem.
Previous research elsewhere that has used participatory decision making to manage predators has sometimes found a lack of common ground between stakeholder groups (Redpath et al., Reference Redpath, Arroyo, Leckie, Bacon, Bayfield, Gutiérrez and Thirgood2004; Johnson & Sciascia, Reference Johnson and Sciascia2013). This lack of agreement could stall management progress or inflame conflict between groups. Namibia is no exception, and the issue is further complicated by divergent stakeholder opinions on managing carnivores (Mosimane et al., Reference Mosimane, McCool, Brown and Ingrebretson2014; Rust, Reference Rust2015a). As many carnivore species range beyond farm boundaries it is essential that farmers manage carnivores collectively.
This study used a novel participatory decision-making exercise to determine (1) whether stakeholders could agree on ways to mitigate human–carnivore conflict on commercial livestock farms in Namibia, and (2) whether there were separate groups of participants who had similar or conflicting viewpoints on preferred management plans.
Methods
The decision-making technique
This study employed an innovative combination of the Delphi technique and Q-methodology. The Delphi technique is a systematic, iterative process of decision making in which experts can form consensus on how to tackle a complex problem (Dalkey, Reference Dalkey1972). The main benefit of using this method is that it is completed anonymously, which can encourage honest discussion, lower inhibitions (Hess & King, Reference Hess and King2002) and reduce power differentials between participants (Dalkey, Reference Dalkey1972). The iterative nature facilitates learning, potentially resulting in a more informed decision (Hung et al., Reference Hung, Altschuld and Lee2008), and can also break down barriers between conflicting stakeholders when they realize that opposing groups potentially hold similar views to their own.
Q-methodology is a structured quantitative interview where participants are asked to rank predefined statements on a scale (Stephenson, Reference Stephenson1953; Brown, Reference Brown1980). It ‘considers people as whole entities and correlates individuals instead of traits’ (Byrd, Reference Byrd2002, p. 52). This technique is particularly suited to studying complex phenomena in which individuals hold contrasting views (Barry & Proops, Reference Barry and Proops1999), and can reveal areas of statistical consensus and disagreement, as well as uncover distinct narratives amongst participants. Consequently, it could be useful in facilitating understanding of heterogeneous stakeholder views on human–wildlife conflict (Johnson & Sciascia, Reference Johnson and Sciascia2013). It does not require a large or random sample size, as participants are chosen based upon reaching theoretical saturation in terms of the possible range of views on a topic (Stephenson, Reference Stephenson1953), and therefore external validity cannot be conferred. However, the aim of Q-methodology is to determine the range rather than the frequency of views (Johnson & Sciascia, Reference Johnson and Sciascia2013).
As the Delphi technique does not seek to address the subjectivity in decision making it could be useful to combine it with Q-methodology. This blend would give rise to a participatory decision-making tool that includes group feedback and repeated rounds, potentially resulting in statistical consensuses and disagreements, and/or various narratives for solving a problem. Deliberation and feedback of results could lead to a more holistic and rational decision, rather than quick, instinctive decisions that might not have considered fully all available options and outcomes (Dalkey, Reference Dalkey1969).
Q-methodology statement collection
Q-methodology requires an initial data collection period to develop the concourse (i.e. the diversity of views on a phenomenon). Content analysis of Namibian newspapers was used to collect some of the concourse statements on the types of mitigation techniques used to reduce conflict with carnivores and livestock farmers in the country. Five of the main English-language newspapers were used in content analysis: Informante, The Namibian, Namibia Economist, Namibian Sun and New Era. Articles were screened by searching for the keywords ‘carnivore’, ‘predator’, ‘human–wildlife conflict’ and ‘depredation’. Content analysis was also used to identify the main stakeholder groups involved in Namibian carnivore management, which informed later sampling for the Delphi/Q.
Interviews were also used to collect additional concourse statements and were conducted with 45 participants: 22 farmers, seven conservationists, five government officials, four tourism operators, three landowners offering trophy hunting, two meat industry workers and two academics. Conducted in English (the official language of Namibia), a semi-structured format was used and interviews typically lasted 1 hour. Questions related to how participants would like to manage carnivores on Namibian commercial farmland, what methods they thought were and were not effective, and what methods they would recommend using.
In total, 50 statements on how to mitigate conflict were collected from the interviews and 78 from newspaper articles. These were refined to 34 by deleting duplicates and combining those that were similar. These 34 statements comprised the concourse for use in the Q-methodology (Table 1), which retained the original wording to capture the intent of the source (Rastogi et al., Reference Rastogi, Hickey, Badola and Hussain2013).
Q-sort statements used during Delphi rounds, with corresponding z scores (underlined values reflect areas of statistical consensus).
*Denotes significant difference between the two factors.
Sampling
As random sampling is not necessary for either Delphi or Q (Brown, Reference Brown1980; Skulmoski et al., Reference Skulmoski, Hartman and Krahn2007), participants were purposefully sampled to ensure breadth and diverse representation across stakeholder groups, to capture the range of possible views on conflict mitigation (Brown, Reference Brown1980). Snowball sampling was used to increase the sample size and ensure theoretical saturation.
A total of 54 potential participants were contacted via email to request their participation in the Q/Delphi exercise. These included all individuals that had participated in the initial interview and those who had been identified through purposeful and snowball sampling. Thirty-five participants (the P-set) completed the first online survey (66% response rate): 14 livestock farmers, six conservationists, six landowners offering trophy hunting, five meat industry employees, two tourism operators and two environmental academics. Emphasis was placed on livestock farmers because they are the stakeholders that currently have the power to manage carnivores on farmland. The same 35 participants were sent the second survey 1 month later, of which 32 completed the survey. Twenty-nine of the 35 completed the third and final survey, of which 11 were farmers, six conservationists, five landowners offering trophy hunting, four meat employees, two tourism operators and one an environmental academic.
Q-sort and Delphi
A Q-sort refers to the participant-ranked concourse statements. An online survey tool (SurveyMonkey, 2015) was used to administer the Q-sort. Prior to implementation, a pilot of the Q/Delphi survey was administered to eight volunteers (four within and four external to the study site) to determine the ease of completing the survey, whether the statements were clear and whether the instructions were comprehensible.
In the survey, participants were asked to rank each mitigation method on a 7-point Likert scale ranging from −3 (strongly disagree) to 3 (strongly agree). At the end of each survey participants were asked about their reasons for strongly agreeing or disagreeing with statements to gain information on the subjective reasoning behind their selections. At the end of the first round, participants were also asked to recommend other statements that should be included in the subsequent rounds, and this resulted in the inclusion of two more statements (35 and 36; Table 1). Each Q-sort was conducted monthly, in three iterations during October–December 2013. At the end of each round an email was sent to participants listing the three most and least popular statements from that round, which comprised part of the Delphi group informed feedback.
Analyses
The Q-sort analysis was run in PQMethod v. 2.33 (Schmolck, Reference Schmolck2002). The analysis determined (1) whether participants reached agreement on any of the proposed mitigation measures and (2) whether there were groups of participants who ranked statements in a statistically similar manner. Initially the data were run through a principal component analysis to identify loaded factors (i.e. groups of participants who ranked statements similarly; Buckley, Reference Buckley2012). These factors went through Varimax rotation, which determined the most parsimonious structure that explained the highest variability between factor groups (Brown, Reference Brown1980). PQMethod arranged the Q-sorts into the factors that were most correlated. Factors with eigenvalues > 1 (Webler et al., Reference Webler, Danielson and Tuler2009) were put into a factor analysis. Each statement was given a z-score based on the mean rank given by participants within each factor. Statistical consensus was defined where P > 0.01 (i.e. groups of participants did not rank statements differently at the 99% confidence level). For the statements that were statistically consensual a strong agreement was defined where the mean rank between factors was at least + 2 or − 2. A grounded theory approach was used to qualitatively analyse answers to the open-ended questions, where common themes were searched for and coded in the data (Strauss & Corbin, Reference Strauss and Corbin1990). Quotes used in the results section were selected for their typical representation of a particular theme (Auerbach & Silverstein, Reference Auerbach and Silverstein2003).
Results
Overall, participants agreed throughout the three Delphi rounds that human–carnivore conflict could be mitigated firstly by training farm workers in effective husbandry to deter predators and secondly by teaching people about conservation of predators (Table 1). Participants also strongly agreed that conflict would not be resolved by reducing consumption of wild game meat.
Despite consensus, Q-methodology separated two factors (or narratives) in each Delphi round (Table 1). Participants in the first narrative (factors A, C, E; Table 1) remained positive regarding non-lethal methods to mitigate conflict (statements 10, 16 and 30) and negative regarding lethal methods and having a government-run livestock breeding centre to replace livestock killed by predators (statements 7, 21, 27 and 34); this group is therefore called the non-lethal narrative. The second narrative (B, D, F) had participants who remained positive regarding training, lethal control and consumptive use (statements 7, 23 and 26) and negative regarding economic incentives and changing from small stock to cattle farming (statements 15, 19, 24, 25 and 32); this group is called the lethal narrative. In general, the non-lethal narrative consisted of conservationists, academics, trophy hunters and cattle farmers, whereas the lethal narrative included sheep farmers and meat industry employees. By the final round, 63% of all livestock farmer participants loaded into the non-lethal narrative.
Factor analysis of Round 1 produced two factors that explained 51% of the variance (38% for factor A and 13% for factor B). The statements, their mean ranking and the areas of agreement and disagreement are shown in Table 1. Participants in factor A (the non-lethal narrative) were defined by agreeing to statements that improved livestock husbandry (statements 4, 13, 30) and disagreeing that predators should only survive in protected areas (statement 34). Conversely, participants in factor B (the lethal narrative) were critical of solutions that involved allowing carnivores to live on farms (statements 24 and 25) and agreed with consumptive use of carnivores (statements 3, 23, 26) but disagreed with changing livestock management practices (statements 13, 22, 33). A strong agreement was reached in Round 1 on statements 11 (training farm workers on how to look after livestock when predators are present) and 16 (teaching people about ecology and the value of predators) as the most acceptable methods to resolve conflict. For statements 18 (reduce wild meat consumption), 25 (pay farmers for the number of predators on their farm) and 27 (have a livestock replacement centre) both factors agreed that these methods would not reduce conflict.
Factor analysis of data from Round 2 produced two factors that explained 45% of the variance (35% for factor C, 10% for factor D). Factor C (the non-lethal narrative) voted in favour of price premiums on predator-friendly meats (statement 4), whereas factor D (the lethal narrative) voted against this. Participants loading onto both factors were more positive about piloting a predator-friendly farm (statement 29) compared with Round 1, although the non-lethal narrative was consistently more positive about this method than the lethal narrative. A strong agreement was reached on the same statements as in Round 1, but views on paying for the number of predators on farms were less negative in this round (statement 25).
Factor analysis of Round 3 again produced two factors that explained 57% of the variance (37% for factor E, 20% for factor F). A strong agreement was reached on statements 11 (training farm workers in more effective husbandry) and 16 (teaching people about conservation and value of predators) and there was strong disagreement with statement 18 (reducing wild meat consumption) as ways to mitigate conflict.
Participants’ reasoning for their answers
The comments made by participants at the end of each round helped to explain the subjective reasoning behind their voting behaviour. Education in predator conservation and training on livestock husbandry were thought by many participants to be the most effective ways to mitigate conflict because ‘only if you know enough about nature you can react against it’ (cattle and sheep farmer CS2). Participants were against the idea of reducing consumption of game meat to increase wild prey for carnivores, believing that ‘wild meat feeds the nation’ (cattle farmer CT9).
In general, participants were critical of compensation payments, as they questioned ‘who will finance compensation for killed livestock? Sounds good, but who has the money and will be willing to administrate it on a sustainable basis?’ (cattle farmer CT3), nor did they like the idea of a government-run livestock replacement centre, as the government ‘will not breed what I want and animals might be less adapted’ (cattle farmer CT6). Furthermore, participants thought that the government should not be involved in managing economic incentives: ‘I do not believe that any (governmental) interventions into private business (i.e. livestock farming) will work out, as administration and control thereof will be too complicated and also some farmers will try to screw the system to earn extra money’ (sheep farmer SF2).
Most participants in the non-lethal narrative, which included many of the farmers, appeared to show some tolerance towards predators. They often mentioned that predators should not be eliminated because they ‘are part of nature and the ecology and have a definitive place therein’ (cattle farmer CT7) and ‘are essential in the food chain’ (cattle farmer CT1). Participants were often wary of economic incentives (‘I do not always agree that money solves the problems’; tourism operator TO2) but believed that conflict could be reduced effectively through education and experience: ‘most problems occur because of lack of knowledge. Only knowledge can improve management’ (cattle farmer CT10). Economic incentives were often considered to be a short-term fix but not a long-term solution.
Conversely, the lethal control narrative believed culling predators was important because carnivores ‘need to be controlled, especially those who kill livestock’ (cattle and sheep farmer CS1). This narrative was critical of changing management practices because they thought this was unfeasible: ‘it is no solution to highly restrict (the way of) livestock farming or make it impossible’ (sheep farmer SF1). Herding was not considered practical because of its ‘major rangeland impacts’ (trophy hunter TH2). The lethal control narrative was in favour of ensuring predators remained only in protected areas: ‘There is enough land in Namibia for predators, why must the commercial farmers also keep them?’ (cattle farmer CT3). Participants in this narrative also thought it made sense to receive income from killing predators, as this could offset the cost of depredation: ‘Reimbursing farmers who have losses due to predators is very good and through trophy [hunting] some funds can be generated’ (meat board employee MB2).
Discussion
The results show that stakeholders reached a strong agreement on some techniques to reduce human–carnivore conflict in Namibia. They agreed that training farm workers to improve their livestock husbandry and teaching people about carnivores could reduce conflict. Farm workers in Namibia tend to be from poor backgrounds and have limited education (Hunter, Reference Hunter2004), and therefore it is possible that educating them about effective husbandry will benefit the situation (Rust, Reference Rust2015b). Previous research has shown that increased knowledge of carnivores and livestock husbandry practices can increase tolerance and reduce livestock depredation on Namibian farms (Marker et al., Reference Marker, Mills and Macdonald2003), thus the Delphi/Q-methodology has successfully highlighted areas of agreement that reflect effective solutions to mitigating human–wildlife conflict.
Along with a consensus on how conflict should be mitigated, there was also agreement on how it should not be managed; i.e. through reducing wild game meat consumption. This statement was suggested initially on the basis that it could increase the availability of wild prey for carnivores and thereby potentially limit livestock depredation (Inskip & Zimmermann, Reference Inskip and Zimmermann2009). However, participants opposed this idea because game meat was seen as a benefit accrued by the majority of Namibians, as most farmers hunt game animals for their own consumption and/or for sale to the general public. Wild meat is also used as part-payment of salaries for farm workers and has significant cultural value (Botha, Reference Botha2005; Karamata, Reference Karamata2006). Thus it may not be culturally or economically feasible to introduce this mitigation method in Namibia.
The results show more areas of consensus between stakeholder groups than other similar predator management studies (e.g. Redpath et al., Reference Redpath, Arroyo, Leckie, Bacon, Bayfield, Gutiérrez and Thirgood2004; Johnson & Sciascia, Reference Johnson and Sciascia2013). It is not clear why this is the case; it may be because of the method used or because conflict between stakeholders is less heated in Namibia compared with other areas. This less intense conflict could be an important factor in creating positive collaborations between stakeholder groups for collectively managing carnivores.
Despite there being areas of agreement, two different viewpoints on how to mitigate negative human–carnivore interactions on livestock farms emerged. However, the divide is not as simple as conservationists preferring non-lethal solutions and farmers preferring lethal control of carnivores. On the contrary, two-thirds of livestock farmer participants opposed lethal control, which contrasts with research findings elsewhere (Selebatso et al., Reference Selebatso, Moe and Swenson2008; Schumann et al., Reference Schumann, Walls and Harley2012). It is unclear why this is the case here, particularly as the survey was undertaken anonymously and therefore there was no pressure on participants to conform to social norms (Dalkey, Reference Dalkey1972). It may be that long-standing education on carnivores in Namibia has slowly improved attitudes and behaviour towards predators (Marker et al., Reference Marker, Mills and Macdonald2003) or that the sample size was too small to notice negative attitudes towards carnivores. This confirms the finding of Chamberlain et al. (Reference Chamberlain, Rutherford and Gibeau2012) that it should not be assumed that individuals within a stakeholder group hold uniform opinions on wildlife management (i.e. that all farmers are anti-carnivore) but rather the differences of opinion are more complex, and consensus can be found between diverse stakeholder groups.
For the lethal control narrative management changes on the farm were not considered to be potential solutions to the conflict. Conforti & Azevedo (Reference Conforti and de Azevedo2003) also found that some farmers were unwilling to improve their husbandry practices to reduce depredation. Farmers in general are risk averse and do not tend to change their management unless absolutely necessary (Binswanger & Sillers, Reference Binswanger and Sillers1983). It may therefore be inappropriate to advise conservative farmers on changing their management.
Economic incentives, particularly those offered by the government, were not a preferred conflict mitigation method as participants feared corruption and incompetence, in accordance with previous findings (Dickman et al., Reference Dickman, Macdonald and Macdonald2011; Rust, Reference Rust2015a). Photographic tourism was not considered to be effective at reducing conflict as participants felt it was too difficult to ensure that guests could view rare and illusive carnivores. There was also concern about having strangers on farms, which interfered with the peace and solitude desired by some farmers. On a deeper level, this suggests that money may not be the sole motivator in decision making; happiness depends on many other values besides monetary wealth (Myers & Diener, Reference Myers and Diener1996).
Using Delphi and Q-methodology together may have helped reduce conflict and power differentials between participants, which usually hinder successful decision making in face-to-face negotiations (Susskind et al., Reference Susskind, van der Wansem and Ciccareli2003). This may have been one of the reasons why more participants reached consensus in this study compared with other studies. Remote methods of decision making may therefore be useful when integrating participant views from conflicting stakeholder groups, such as is usually the case in human–wildlife conflict situations. However, participating remotely might have hampered social learning (Ziglio, Reference Ziglio, Adler and Ziglio1996), emotional attachment and empathy building, which are important aspects of decision making (Wieczorek Hudenko, Reference Wieczorek Hudenko2012). I therefore recommend that participants attend a workshop at the final round to facilitate communication between participants.
Undertaking the Q/Delphi online in English could have limited the number of participants involved, but most Namibian commercial farmers are well educated (at least half have a university degree; Olbrich et al., Reference Olbrich, Quaas and Baumgärtner2012). They are also required to use computers as part of record keeping and financial management, and therefore they are usually computer literate and have access to the Internet (NR, pers. obs.). A future study could include surveys in various languages and via other media to ensure participation is as inclusive as possible. Although this method was relatively inexpensive to conduct (the online survey and content analysis were free to run), the process of collecting concourse statements and conducting the three Delphi rounds was time-consuming (6 months in total). If decisions need to be made quickly, this method may not be appropriate. The final limitation to this method is that, because of the sampling procedure used, results cannot be inferred more widely. If this method were to be used to inform policy, a larger sample size using random sampling would be needed.
In summary, combining Delphi and Q-methodology to understand whether stakeholders could agree on how to mitigate carnivore conflict on commercial farms in Namibia revealed some consensus regarding potential management policies, as well as areas of disagreement. This novel method could be used in other areas of participatory decision making for wildlife management, to legitimize the process and reduce conflicts between groups. The fact that the technique required participants to interact remotely and anonymously was probably a key factor in ensuring that conflict between stakeholders did not hinder the process. It may therefore be important for participants to reach agreement remotely, at least during the first few rounds. The suggested mitigation methods can be used as a starting point to help develop a socially accepted carnivore management plan that will assist in reducing conflict between people and carnivores in Namibia.
Acknowledgements
This work was sponsored by the Economic & Social Research Council (grant number 12906529), Panthera Kaplan Graduate Award and Okonjima Nature Reserve/AfriCat Foundation. I thank especially H. & B. Sohrada and S. Redpath.
Biographical sketch
Niki Rust's research focuses on carnivore conservation in unprotected areas and uses social science to understand how to improve coexistence between people and carnivores.
Open access
