Introduction
Conflicts between wildlife and agriculture are increasingly challenging agriculture and wildlife agencies (Krebs et al. 1998, Miller et al. 2013, Miller and Sweeney 2013). Policy to address human-wildlife conflicts is often controversial (Messmer 2009). Developing policy to manage interactions between wildlife and agriculture has been identified as critically important (Jones et al. 2013), yet there remains little research on the societal factors that bring these issues to the policy making agenda for wildlife-agricultural conflicts. The policy process literature has identified drivers of policy creation such as problem severity, interest group involvement, media coverage, and public perception as predictive of the passage or modification of policies to address a social problem (Gilliam Jr and Iyengar 2000, Soroka 2003, Walgrave et al. 2008, Baumgartner and Jones 2010).
Wildlife-agriculture conflict and policy development is often exacerbated by invasive or exotic animals (Pimental 2007). In the United States (U.S.) there are at least 30 species of exotic free-ranging mammals which have become established since European colonization, causing an estimated $46 billion in damage annually (McKnight 1964, Mayer and Brisbin 1991, Pimental 2007). Wild swine (Sus scrofa), which include wild swine (Sus scrofa domestica), Eurasian Russian boar (Sus scrofa linnaeus), and hybrids between the two, are the most abundant free-ranging, exotic ungulate in the U.S. and annually cause an estimated USD$1.5 billion in damage (Pimental 2007, Bevins et al. 2014). Since the 1960s wild swine have expanded their range to at least 38 States and 3 provinces in Canada and continue to increase (Bevins et al. 2014, Brook and Beest 2014). This range expansion has contributed to the impact of wild swine on ecosystems and both livestock and crop agricultural systems in North America (Bevins et al. 2014). However economic impacts and perceived problem severity often differ regionally and by agricultural commodity. For example annual crop damage in California is estimated to be USD$1.7 million while in Georgia it is estimated to be at least USD$57 million annually (Frederick 1998, Mengak 2012). In contrast the economic impacts associated with a cattle-wild swine outbreak of foot and mouth disease (FMD) in the United States have been estimated to be has high as USD$14 billion (Paarlberg et al. 2002). Despite the economic costs wild swine is also valued as a hunting resource and in many states is managed as a wildlife resource.
Wild swine impacts are often recognized at a local or regional scale, with national scale policy development only recently evident. In addition, there is a diversity of public attitudes toward wild swine encompassing, agricultural pest, disease hazard, commodity, source of income, and recreational resource (Tisdell 1982, Izac and O\’Brien 1991). Izac and O\’Brien (1991) found that these perceptions changed with location, time and individual. The combination of public attitude heterogeneity in the regions with wild swine and a neutral or undefined policy image – how a problem and its solution is defined, understood, and discussed – may contribute to the diversity of management positions on wild swine damage control and mitigation. Schattschneider (1960) noted that the essence of policy conflict over a public issue is the scope of participation and the importance of policy image in defining the problem and the solutions. Kingdon and Thurber (1984) found that identifying the visible, and presumably most affected, participants in a policy issue is central to understanding the dynamics of agenda setting and the resulting policy development. Furthermore it is traditionally accepted that problem severity is a significant stimulus for the adoption of policy innovations (Sapat 2004). In the case of wild swine, the primary interest group impacted by increasing problem severity is agriculture (Bevins et al. 2014). However, public perception concerning an issue is also important in prioritizing the policy agenda (Baumgartner and Jones 2010).
Media coverage of public issues – both quantity and tone – has been widely recognized as an important driver in shaping national public perception and policy agendas (Gilliam Jr and Iyengar 2000, Walgrave et al. 2008, Baumgartner and Jones 2010). Media coverage is generally thought to influence policy agendas in two primary ways. First, media coverage can influence the relative salience (importance or prominence) of a particular pubic issue through repeated coverage over time (Soroka 2003, Baumgartner and Jones 2010). Second, media coverage can influence public and policy conceptualization about an issue and coalescence – how an issue is understood, defined, and framed (Elder and Cobb 1983). This conceptualization of an issue can influence the perception of the possible solutions and the importance of addressing the problem with governmental policy (Weart 1988, Baumgartner and Jones 2010). However, there is mixed evidence for how these factors – problem severity, interest groups, media coverage, and public perception – may act together to influence policy generation (McCombs and Shaw 1972, Funkhouser and Shaw 1990, Entman 1993, Koch-Baumgarten and Voltmer 2010).
Our objectives were to characterize the relative influence of the factors that led to the establishment of the APHIS National Feral Swine Damage Management Program in 2013 (federal government fiscal year 2014). Specifically we wanted to understand 1) the significance of public policy image on congressional policy activity; 2) to assess the influence of problem severity and broad governmental institutional pressures associated with expansion of wild swine at a national level on policy activity; and 3) to identify predictors of policy activity for informing wildlife and agricultural interface management; specifically program assessments and new program development. Here we use the term ‘policy’ in its broadest definition referring not only to operational policies of government but also including all dialogue related to the development of policy. To investigate the relationship between policies, wild swine, and agriculture we use 29 years of data from three primary datasets – number of wild swine related policy actions (response variable), newspaper headline data, and the amount of agriculture in wild swine regions. Based on studies suggesting a strong dependence of policy change on changes in public policy image (Jones and Baumgartner 2004, Baumgartner and Jones 2010), specifically increased policy activity when public policy images become negative, we hypothesized that significant increase in the number of negative newspaper articles would act as a mechanism for influencing policy activity and provide a link between changes in policy and expanding wild swine populations. Because governmental institutions tend to increase stability in policy areas (Jones et al. 2003, Baumgartner and Jones 2010), we hypothesized that increasing the amount of agriculture in wild swine regions might be related to increasing problem severity and result in increased pressures on federal governmental institutions. Thus increasing policy activity as agricultural related interests increased demands for policy solutions to wild swine related issues – agricultural damage and economic losses. In our statistical models, we wanted to estimate these effects and determine if these patterns are consistent with increased policy activity. The broader goal of this analysis is to provide a mechanistic understanding of the policy image and institutional conditions that give rise to variations in the policy process, which enables improved response to changes in conditions that impact both wildlife and agricultural policy.
Methods
Congressional Policy Action Data
A systematic search of the United States Government Printing Office Federal Digital System (FDsys) (GPO 2014) was used to generate data describing congressional activity related to wild swine. FDsys is an official repository of all official publications from all three branches of the United States Federal Government and currently contains over 7.4 million electronic documents from 1969 to present. Our search included congressional hearings, congressional record, congressional reports, bills, and changes to the code of federal regulations from 1985 until 2013 when the APHIS National Feral Swine Damage Management Program was established. Documents included in our study contained any of the following terms: ‘feral swine’, ‘feral hog’, or ‘feral pig’, ‘wild swine’, ‘wild hog’, or ‘wild pig’. Each document was considered an independent policy action, and the number of documents by year was tallied to generate count data by document type, primary agricultural commodity (livestock or crop) the document addressed, and year. Our method may have included documents which were not specifically addressing wild swine related policy; to evaluate this assumption a 5% random sample was taken and the documents were classified as addressing wild swine related policy or not. Based on the results of this assessment we assumed that if the document contained reference to wild swine the issue of wild swine was either on the policy agenda or influencing the agenda in some way.
Media Data
To generate data on media reporting of wild swine related topics a systematic search of four major news consolidators was performed – Newsbank, LexisNexis, EBSCO, and ProQuest (EBSCO 2016, LexisNexis 2016, NewsBank 2016, ProQuest 2016). Our review was restricted to newspaper articles published from 1985 to 2013 in the United States. In order for an article to be included it must have contained the terms ‘feral swine’, ‘feral hog’, or ‘feral pig’, ‘wild swine’, ‘wild hog’, or ‘wild pig’ in the title or lead in to the article. Articles published by the same media source and author on the same date were considered duplicates and removed. The data were summarized generating three annual predictors, the number of articles, the number of different media sources, and the number of states with at least one article.
Each article headline was classified as positive or negative. Our assumption here was that the article headline summarized the overall content, or conclusion of the article. In order to classify articles as having positive or negative tone we used a polarity index described by Rinker (2013) and Breen (2012). In general this polarity algorithm uses a word sentiment (positive or negative) dictionary (Hu and Liu 2004) to tag polarized words in the article headline. A context cluster of six words is extracted from around each polarized word (positive / negative) in the article. The words in this cluster are identified as neutral, negator, amplifier, or de-amplifier. Neutral words hold no value but do affect word count, while each polarized word is counted and weighted in the context cluster. The context clusters for the article headline are summed and divided by the square root of the word count yielding an unbounded score for article describing the negative or positive tone of the headline.
For our purposes we are interested in the cumulative influence of article tone and media sources. In order to produce a measure of this annual cumulative article tone we generated the annual mean tone. This was then multiplied by the number of articles published in the year and by the number of sources creating two predictor variables describing the annual tone for media sources (source tone) and the annual tone for articles (article tone). Classification of newspaper headlines and generation of the media tone indices were done using the qdap qualitative data and quantitative analysis package (Rinker 2013) within the R computing environment (RCoreTeam 2016).
Agriculture in Wild Swine Affected Regions
To generate a measure of the amount of agriculture present in regions (defined as counties) with wild swine we compiled two data sources. National Agricultural Statistics Service (NASS) data reporting the number of farm operations present in each U.S. County were used as a measure of all farms (USDA 2014). An aggregate of data describing the distribution (presence/absence) of wild swine at the county level was compiled from the Southeast Wildlife Disease Cooperative Study (SCWDS) (SCWDS 2013) and two publications Waithman et al. (1999) and Hanson and Karstad (1959). These data represent the known county level distribution of wild swine over the past 50 years (Figure 2.5). These data were merged with NASS data describing the number of farm operations to generate a national level measure of the proportion of farm operations in counties where wild swine were known to occur.
Formulation of competing models
Co-occurrence of Wild swine and Agriculture Models
Because data describing the distribution of wild swine are not available for all years and represent samples of the known distribution of wild swine over time, models were fit to these data to estimate the national proportional change in the number of farms in counties where wild swine occur for each year from 1959 to 2013. This allowed estimation of the expected proportion of farms co-occurring with wild swine in years without data. We determined relative support in the data for four candidate models – linear, exponential, power, and logistic – to describe the phenomenological change in national wild swine-agriculture co-occurrence. The best approximating model was used to represent the proportion of agriculture in regions with wild swine for each year and was used as a predictor in the policy models.
Policy Models
We evaluated support for competing models portraying the relationship between the annual count of policy actions (response) and six variables of interest measuring annually the 1) number of newspaper articles, 2) number of news sources, 3) number of states with newspaper articles, 4) negative tone for news sources 5) negative tone for newspaper articles, 6) and the proportion agriculture in regions with wild swine, here on referred to as agriculture. Specifically, these independent variables represent hypotheses about specific mechanisms that resulted in congressional policy activity that eventually resulted in the establishment of a National program to address the problem.
- Relative Salience: An increase in the number of newspaper articles, media sources, and the number of states with media reporting would increase the salience of the policy image increasing congressional policy actions.
- Problem Coalescence: An increase in negative media tone for wild swine represents coalescence of the policy image increasing congressional policy actions.
- Institutional Pressures: Increasing the amount of agriculture in wild swine regions is related to increasing problem severity and results in increased pressures on Federal government institutions to find a policy solution thus increasing Federal congressional policy activity.
These processes act as surrogates to capture important policy related mechanisms. The increase in the number of negative newspaper articles indicates a change in the policy image (salience and coalescence) and the increase in the amount of agriculture potentially impacted, that is associated with the geographic expansion of wild swine, indicates a change in problem severity and federal government institutional policy pressure (Elder and Cobb 1983, Jones et al. 2003, Soroka 2003, Sapat 2004, Baumgartner and Jones 2010).
Model selection
We used multi-model inference within an information-theoretic framework, (Burnham and Anderson 2002, Burnham et al. 2011) to estimate model parameters describing the probability of congressional policy actions related to wild swine. All models assumed a Poisson error structure and were fit using a generalized linear model with a log link function. Akaike information criterion with a correction for small sample size (AICc) was used to assess the relative information content of the models. We fit all subsets of the global model and computed model-averaged regression coefficients, unconditional standard errors (SE), cumulative AICc weights of evidence as a measure of variable importance (Burnham and Anderson 2002, 2004, Burnham et al. 2011), and 95% confidence intervals (Burnham and Anderson 2002, 2004). We used a shrinkage estimation approach to produce unconditional model averaged parameter estimates, in which covariates that did not appear in a particular model subset were assigned coefficients of zero to avoid biasing coefficient estimates away from zero (Burnham and Anderson 2002). Our interpretation of the explanatory power of the regression coefficients in our model was guided by three measures: 1) the weights of evidence, ranging from 0 to 1.0, where higher weights indicated greater relative importance; 2) the 95% confidence interval for each regression coefficient that did not overlap zero; and 3) effect sizes indicated by each regression coefficient. The final inferential model was used to estimate the relative annual contribution of each predictor to policy activity across the 29 years investigated and to estimate the relative contribution of livestock and crop agriculture to annual policy activity for wild swine. Maximum likelihood estimates, confidence intervals on model parameters, and AICc values were obtained using MuMIn Multi-Model Inference package (Barton and Barton 2015) available in R (RCoreTeam 2016).
Model validation
AICc does not represent a goodness-of-fit metric hence we assessed model fit using k-fold cross-validation which contrasts the number of policy actions predicted by the model and the observed frequency of policy actions (Kohavi 1995). Based on Huberty’s rule (Huberty 1994), we first randomly divided the wild swine policy action data among four cross-validation folds. We used each possible set of three folds to fit a predictive model, again employing multi-model averaging, which we then used to predict the fourth withheld fold. Results of 100 iterations of this process, each with a new random allocation of data across four cross-validation folds, were averaged to avoid dependency of validation results on a single random allocation of data across folds. We binned predicted values from our cross validation results and then calculated a Pearson correlation between those values and the observed number of policy actions within each bin. We assessed the performance of our final model using the Pearson correlation which provides a more rigorous measure of the linear agreement between predicted and observed policy activity. Because validation results can be sensitive to binning method (Boyce et al. 2002), we applied and compared the results using a quantile binning method for 4, 10 and 20 bins. Cross-validation was implemented in R statistical software (RCoreTeam 2016).
Results
Congressional Policy Actions
Our search of policy documents identified 421 documents related to wild swine. Figure 2.6 presents the distribution of these documents by type along with key milestones in the emergence of the wild swine policy area. Evaluation of a random sample of 22 (5%) of these documents to determine if the assumption that documents containing a reference to wild swine were an indicator of wild swine policy activity found that all (100%) were related to wild swine policy. This indicated that our assumption was valid and the policy document frequency data represented policy activity related to wild swine. Assessment of these documents identified roughly four policy periods of increasing policy activity – no activity, regulatory, hearings, and implementation. The period from 1985 to 1993 showed no observable policy activity. This was followed by a brief period from 1994 to 1998 of changes to the federal register and the code of federal regulations (i.e. regulatory activity). From 1999 to 2007 in addition to regulatory activity, discourse on feral swine began in the form of congressional hearings. The last stage was dominated by policy implementation from 2008 to 2013 which accounted for 63.9% of the total activity and comprised both regulatory and distributive policies. Across all years the policy actions largely represented agricultural related issues (68.8%) and were dominated by concerns associated with livestock agriculture (46.1%). In general there is a rapid increase in activity related to wild swine, starting with the minor regulatory changes, followed by congressional hearings, then policy implementation, and in 2013 the establishment of a new national program to address wild swine damage.
Media Data
We identified 1,016 unique newspaper articles related to wild swine between 1985 and 2013. Figure 2.7 illustrates the media data from 1985 through 2013. As illustrated, the number of newspaper articles, number of media sources and number of states with newspaper articles were relatively constant prior to 1998 with a rapid increase in articles, sources, and states after 1999. This period from 1999 to 2013 accounted for 95.7% of articles and 84.8% of news sources. The number of states with wild swine related media reports continued to increase throughout the study period with 45 states having at least one article.
Analysis of newspaper article tone found that mean tone was close to neutral for both number of media sources (-0.40±1.39) and number of articles (-0.25±1.55) from 1985 to 1998. Figure 2.7 also presents the change in media tone over time. From 1999 to 2006 polarity became increasingly negative for both the media sources (-3.77±2.43) and number of newspaper articles (-3.77±2.38). During the implementation period from 2007 to 2013 polarity continued to become negative for media sources (-13.14±5.82) and number of newspaper articles (-17.78±4.58).
Co-occurrence of Wild Swine and Agriculture Models
The co-occurrence of wild swine and agricultural operations expanded at an increasing rate from 1959 until 2013. Based on the AICc weights the best approximating model was a logistic model (Table 2.1). For our study period the proportion of farms in wild swine regions increased from 0.17 in 1985 to 0.41 in 2013 (Figure 2.8). This represented an annual rate of increase of 1.01 (stdev <0.01) during this period. The estimated inflection year was 2034 with 69.9% of farms in regions with wild swine. Based on the strong predictive capacity of this distribution (Adjusted R2 = 0.99) it was used as a predictor in the policy models to represent the number of farm operations potentially impacted by wild swine annually which we use as a surrogate for changes in institutional policy pressure.
Policy Models
Based on the final inferential model policy activity was most strongly associated with the number of states with newspaper articles, polarity of media sources, polarity of newspaper articles, and the proportion of agriculture in wild swine regions (Table 3). Covariates representing each of these four factors had high AICc weights of evidence and 95% confidence intervals that did not include zero indicating high predictive importance. Cross-validation indicated that the final model had strong predictive capacity. The quantile binning method produced similar Pearson correlations of 0.969 (4 bins), 0.915 (10 bins) and 0.957 (20 bins) between median predicted policy actions and the observed policy actions in each bin, indicating low sensitivity
Parameter estimates and odds ratios for the parameters considered are shown of the cross-validation results to binning method in Table 2.3. The number of states with wild swine related newspaper articles was a positive predictor of wild swine policy activity (odds ratio = 2.08). For every additional 5 states with newspaper headlines related to wild swine there was a 3.65% increase in the number of policy actions. Increasing negative tone of both number of newspaper articles (odds ratio = 1.95) and number of media sources (odds ratio = 1.14), increased in the number of policy actions. That is for every 10 negative newspaper articles and 10 additional negative media sources wild swine policy activity increased by 6.7% and 1.3%. The proportion of agriculture in regions with wild swine was the most significant predictor of policy actions (odds ratio = 4.09); that is for every 1% increase in the proportion of agriculture in regions with wild swine policy activity increases by 41%. This increase appeared most sensitive to livestock related policy activity. Livestock policy activity (41%) increased at nearly twice the rate of crop policy activity (23%) for every 1% increase in the proportion of agriculture in wild swine regions and was also a significant predictor of livestock (odds ratio = 4.08) and crop (odds ratio = 3.43) specific policy activity for wild swine. Figure 2.9. illustrates the functional relationship between increasing agriculture in wild swine regions and the resulting change in all wild swine policy activity and policy activity specific to livestock and crop agriculture.
The contribution of the four most important predictors, number of states with newspaper articles, polarity of media sources, polarity of newspaper articles, and national proportion of agriculture in wild swine regions, to policy activity changed across the 29 years evaluated (Figure 2.10). The annual contribution of the proportion of agriculture in wild swine regions to policy activity varied the most, with a 54.9% change from 5.5% of policy activity in 1985 to 60.7% of policy activity in 2012. Both media source and newspaper article tone had declining annual contribution to policy activity, declining 37.5% and 17.2%. Combined media source and newspaper article polarity contributed to 30.5% of policy activity in 2013 compared to a combined 71.7% in 1985. The number of states with newspaper articles contributed a consistent amount annually (mean=22.8%; 95% CI = 21.1%-124.5%) to policy activity across all years.
Discussion
Our models found a linkage between policy activity and four predictors representing number of states with media, media tone and agriculture. These predictors have been described in previous studies as representing specific policy processes associated with policy image salience (number of states with news articles) (Gilliam Jr and Iyengar 2000, Schnell 2001, Soroka 2003, Walgrave et al. 2008, Baumgartner and Jones 2010), policy image coalescence (newspaper article and source tone) (Elder and Cobb 1983, Weart 1988, Baumgartner and Jones 2010) and institutional pressures (feral swine-agriculture co-occurrence) (Kingdon and Thurber 1984, Sapat 2004). Further we found the contribution of these predictors to policy activity changed across the 29 years analyzed indicating the development of federal feral swine policy went through a continuum of policy development. Understanding how these predictors that serve as proxy measures of policy stages contribute to policy development can provide a better understanding of important latent processes that give rise to national policies to address wildlife problems. This in turn can support the development of programs and policies that best address the social issues. Here we provide a discussion of these predictors and their potential contribution to the latent policy process that may link the observed policy activity with social and institutional processes.
Our model suggests that for wild swine policy, increasing negative newspaper articles may have acted as a mechanism for influencing initial policy activity. This may have been particularly important for issue emergence and salience as media related predictors contributed most during early phases of policy development. In addition our results suggest that increasing the amount of agriculture in wild swine regions influenced policy activity the most, particularly during later policy phases of issue coalescence and policy implementation. Media predictors for newspaper article and media source tone became less important once policy implementation began, indicating that the introduction of potential policy solutions by governmental institutions may have increased stability in the policy area (Jones et al. 2003, Baumgartner and Jones 2010).
The emergence of wild swine as a policy issue was characterized by decades of general inattention and no observed policy development (see Figure 2.6). The lack of policy attention and media coverage that was neutral indicates that wild swine was not a broad issue prioritized by society nor did the issue have a distinct policy image prior to 1994. This lack of policy image may have contributed to policy inattention despite recognition in the scientific literature that wild swine were damaging to both the environment and agriculture (Hanson and Karstad 1959, Wood and Barrett 1979). Policy images have implication for which interest groups and governmental institutions become involved, how an issue is understood, and which institutional venue an issue will be addressed by government (Baumgartner and Jones 1991). The scope of participation by stakeholders and the clarity of a policy image is central to defining the policy agenda and potential policy solutions (Schattschneider 1960). However, public perception concerning an issue is also important in prioritizing the policy agenda (Baumgartner and Jones 2010). In the case of wild swine without a distinct policy image prior to 1999 the primary agricultural interests – livestock and crop agriculture – may not have been able to pursue a policy solution at the national level and as a result did not contribute significantly to policy activity prior to 2005. Furthermore, the lack of public attention and limited access by the primary interests may have limited the issue from being considered on the policy agenda.
National policy issues of wild swine appear to have emerged sometime after 1994 when the first policy activity for wild swine occurred. The five years from 1994 to 1998 had a nearly six fold increase in news coverage. However the news coverage tone during this period was not significantly different than neutral, indicating that the general public was aware of the emerging issue but there was no consensus, and the issue had not yet become salient. The lack of salience during this time may also be evidenced by the continued importance of news predictors relative to proportion agriculture in regions with swine (Figure 2.10). Previous studies (Jones and Baumgartner 2004, Baumgartner and Jones 2010) have proposed that increasing news media, specifically negative news media, indicates public policy image coalescence and policy issue salience. Prior to 2005 the relative contribution of agriculture to policy activity was less than media predictors indicating a lack of issue salience.
As news media became increasingly negative there was a rapid increase in total policy activity and the relative contribution of agricultural interests to policy activity increased. This rapid change may indicate that the issue became broadly salient and the policy image coalesced around this period. Salience of social issues in public discourse may determine whether or not issues expand on the government agenda (Koch-Baumgarten and Voltmer 2010). For example issue salience can determine voter turnout and choice preferences (Becker 1977). In our analysis news media may have provided a method for establishing issue salience and coalescence, serving to bring the issue to the governmental agenda. In addition, the number of states with negative media was important in all models. However, the relative contribution to policy activity was nearly the same across all years. This may indicate that the geographic distribution of media was important across the entire continuum of policy development but was not related to any one phase of the policy process. Previous authors have proposed that news media can serve as an agenda-setting mechanism (Scheufele and Tewksbury 2007).
Social problems that receive media attention do not automatically receive attention on the policy agenda or generate policy actions. The policy image – how a problem and its solution is defined, understood, and discussed – must be established (Baumgartner and Jones 2010). In the case of wild swine, the policy image does not appear to have begun to become broadly salient and coalesce until the 2000s when media tone became significantly negative. The first congressional hearing addressing wild swine was conducted in 1999 and addressed issues related to U.S. Department of Agriculture’s (USDA) policy for addressing wildlife transmission of diseases to domestic livestock and specifically brucellosis in wild swine (Senate 1999). This has been identified as a potentially significant issue facing agriculture and wildlife management (Miller et al. 2013). However congressional hearings did not begin in earnest until 2005 and 2006 when ten hearings were held – over double from the previous five years. Hearings in these two years were largely related to potential animal agricultural impacts associated with classical swine fever, a swine disease with international trade implications for the U.S. swine industry (Paarlberg et al. 2009).
Once the issue of feral swine was on the policy agenda, we found that the presence of agriculture in wild swine regions, a proxy for institutional pressures on policy makers, had a greater influence on the overall frequency of policy activity (Figure 2.10). The relative contribution to policy activity shifted from primarily media related to primarily agricultural interests related sometime after 2005. This indicates that agricultural interests in wild swine regions may have been the primary driver in bringing wild swine to the institutional agenda. In addition the focus of wild swine related policy activity on agricultural damage related issues indicates that agricultural interests had influence in setting the agenda. Further, livestock agricultural interests appear to have dominated the problem definition period, contributing to as much as 80% of all policy activity prior to 2005 (see top panel in Figure 2.7). Previous studies have proposed that actors which are able to define the problem early in the issue emergence stage tend to control future policy development even if new actors inter the policy arena (Schattschneider 1960).
A policy image serves to link the problem with the governmental solution (Baumgartner and Jones 2010). The coalescence of a policy image of wild swine negatively impacting agriculture was succeeded by a flurry of policy activity. Between 2007 and 2013 there were an average of 38 wild swine related policy actions a year with 67% of these directly related to implementing policy, rulemaking, or appropriations. This included the Feral Swine Eradication and Control Pilot Program Acts of 2009 and 2011 that authorized the secretary of the Department of Interior (DOI) to provide financial assistance to specific states for eradication efforts (Landrieu and Vitter 2009, Landrieu 2011). This was followed by the establishment of the USDA Animal Plant Health Inspection Service (APHIS) National Feral Swine Damage Management Program in 2013 (USDA 2013). The focus of policy activity during this policy implementation stage remained dominated by agriculture (69.1%), specifically livestock disease related concerns (46.1%). Crop agricultural concerns remained much lower with only 23% of activity related to crop damage. This period also represented the greatest number of negative newspaper articles, which occurred at a rate of 107 per year in 27 states. The proportion of agriculture in regions with wild swine continued to increase with an estimated 38% of farms in the United States potentially impacted. Our model indicates that both media and livestock agriculture were predictive of increased policy activity.
Increasing, problem severity and the resulting institutional pressures by agricultural interests to address agricultural damage likely contributed most to the development of policy. This is evidenced by the proportion of agriculture in regions with wild swine being the most significant predictor of wild swine policy actions in our model (odds ratio = 4.09). Furthermore, agricultural interests accounted for over two-thirds of all wild swine related policy activity and nearly dominated policy activity (~95%) during the issue emergence stage. Our analysis also suggested the emergence of a policy subsystem with relatively few actors dominated by agricultural interests and particularly livestock agriculture (Heichel 1990). Livestock agriculture had nearly twice the influence on wild swine policy activity and likely contributed most to forming the policy image supporting the notion of a policy subsystem. The interaction of beliefs and values concerning a particular policy with the existing set of institutions, in this case USDA and DOI, acted as the venues of policy action (Baumgartner and Jones 1991). This often results in policy subsystems that are oriented towards a given industry (Baumgartner and Jones 2002a). Furthermore interactions of venue and image can produce self-reinforcing system characterized by a positive feedback that tends to create stability over time (Baumgartner and Jones 2002b). In the case of wild swine this may represent a relatively stable policy subsystem devoted to wild swine control and agricultural damage mitigation.
This study is based on a large search of government documents and media data; therefore there are inherent constraints on inference. While our objective was to investigate the relative contribution of media and institutional pressures on national wildlife-agricultural policy development, there are other potential drivers of policy activity. Previous studies have found that interest group access to congressional committees and advisory committees are influential in the development of policy (Balla and Wright 2001), although this is also influenced by the number of stakeholders in a policy area (Baumgartner et al. 2009). In our study we only considered three actors – livestock agriculture, crop agriculture, and the public – although there may have been additional actors that contributed to the generation of national policy. We also did not consider other potential processes that may have influenced national policy activity such as policy diffusion (Berry and Berry 1999) or policy entrepreneurs (Mintrom and Norman 2009). These policy processes may also have contributed to the observed policy activity. While our study provides insights into drivers of policy activity addressing the wildlife-livestock interface, it could be enhanced by investigating these other mechanisms that may also be important in creation of policy.
Conclusion
News media coverage has been identified in previous studies to influence policy development (Gilliam Jr and Iyengar 2000, Schnell 2001, Soroka 2003, Walgrave et al. 2008, Baumgartner and Jones 2010). However, we found that institutional pressures applied by actors may have a far greater contribution to the development of policy. This may be particularly important when economic impacts are a result of the issue. This effect maybe even greater early in the emergence of an issue when fewer actors are involved and the ability for actors to define the problem and the eventual policy image is greater (Schattschneider 1960). Furthermore, among potential interests involved during this stage, those with the greatest potential risk for damage may have the greatest impact on the formation of policy and in the case of wild swine this appears to be livestock agriculture (Baumgartner et al. 2009).
This is the first analysis we are aware of that examines the role of public sentiment and institutional pressures on the development of policies that address the wildlife-livestock interface. In the case of wild swine, our model suggests that changes in co-occurrence of wild swine and agriculture over the last 29 years, resulting in increased problem severity (increased agricultural damage), likely contributed most to the eventual development of policy to mitigate the issue. This likely resulted from increasing industry pressure on agricultural agencies to protect or mitigate damage associated with wild swine. This was evidenced by our model results and also the significant consideration given to agricultural damage caused by wild swine during congressional hearings and in congressional reports (GPO 2001, 2013). Congressional hearings and reports often focused on specific mechanisms, rules, or procedures the USDA had in place to mitigate livestock and crop damage caused by wild swine (GPO 2013). In addition the livestock agricultural sector likely contributed more to the development of policy with nearly half of policy addressing livestock related issues, particularly the potential for disease outbreaks. This may be driven by the potential for large economic losses – USD$5.8 billion – associated with a single livestock-wild swine disease outbreak (foot and mouth disease) compared with the currently estimated USD$800 million in damage caused by wild swine to crop agriculture (Paarlberg et al. 2002, Pimentel et al. 2002).
Given the scarcity of rigorous quantitative policy work in this system – wildlife-agriculture interface, specifically wild swine – greater attention is needed to disentangle the mechanisms driving policy development. Although work has been conducted examining the influence of public sentiment and news tone influence on policy development (Baumgartner and Jones 2010), there remains a lack of information linking measures of public perception and institutional pressures specifically for the wildlife-agricultural interface. Such information could provide valuable insight into the observed variability in policy approaches addressing wildlife-agriculture interactions. Unravelling the role of news media coverage, public knowledge, public sentiment, agricultural impacts, and environmental damage—as it relates to changes and differences in policy approaches at local and national levels—will require large-scale studies on local and national policy drivers which have not been attempted to date for the wildlife-livestock interface; but could provide valuable information for improving policy systems for control and mitigation of damage associated with wild swine and wildlife in the United States. Policy makers can in turn use analyses such as this to better design policies that align with public interests and benefactors ensuring long term success of policies by incorporating all interests (Loomis and Helfand 2001).
2017-6-14-1497470328