Mapping social conflicts to enhance the integrated management of white‐tailed deer (Odocoileus virginianus)

Understanding the social feasibility of wildlife conservation approaches is essential to reducing social conflicts over wildlife and public backlash toward wildlife agencies and organizations. The Potential for Conflict Index2 (PCI2) and geospatial analyses of conflict can help wildlife practitioners strategically engage their publics, but these two tools have yet to be combined. Using data from a 2021 survey about white‐tailed deer in Indiana (n = 1806), we analyzed conflict levels among stakeholder self‐identities and political ideologies regarding the acceptability of six possible management methods, three lethal and three nonlethal. We then conducted a hotspot analysis of gridded PCI2 values to map areas of high and low social conflicts across the state. Conflict potentials showed more consistent covariation with political ideologies than with stakeholder self‐identities, aligning with urban–rural divides in wildlife experiences. Data on political leanings and residency may thus be more reliable than stakeholder categories to predict social conflicts over wildlife management. Hotspots of conflict over lethal methods clustered around urban areas, indicating that agencies should focus on engaging urban residents about deer management. Our conflict hotspots can be combined with other spatial data to create social units of analysis, which can help practitioners develop targeted and socially accepted strategies for wildlife conservation and management.


| INTRODUCTION
With increasing human populations, deteriorating habitats, globalization, and climate change, human-wildlife interactions are becoming increasingly complex (Abrahms et al., 2023;Dietsch et al., 2016;Harris et al., 2023;Manfredo et al., 2020;Stinchcomb et al., 2022a).Research has shown that human-wildlife interactions are shaped by various social factors and that human-wildlife conflicts "are often manifestations of underlying human-human conflicts" (Dickman, 2010, p. 458).Wildlife agencies and organizations across the globe, including in North America, are striving to incorporate more diverse social values and perspectives into management planning (AFWA & WMI, 2019;Hodgson et al., 2020;König et al., 2021).Scholars have proposed the integration of responsibility, equity, justice, and inclusion as governing principles for human-wildlife interactions (Harris et al., 2023).Specifically, addressing the needs of all possible 'beneficiaries' of wildlife remains an important aspect of good wildlife governance (Decker et al., 2016;Hare et al., 2017).Although scholars posit that understanding and integrating local knowledge into decision-making are essential for expanding support for wildlife management (Armitage et al., 2011;Assche et al., 2017;Bélisle et al., 2018;Robbins, 2006), most wildlife agencies and organizations still face limited social science capacities, narrow funding streams, and embedded institutional cultures that hinder close and consistent engagement with a diverse public (Jacobson et al., 2010;Pomeranz et al., 2021).Thus, there have been calls for wildlife agencies and organizations to undergo sweeping institutional reforms to effectively achieve collaborative and participatory ideals of wildlife conservation and governance (Pomeranz et al., 2021).
Amid institutional change, survey-based research provides crucial tools for wildlife practitioners to begin assessing social diversity and integrating public needs into the planning process.Surveys are widely used among wildlife agencies and organizations to assess public perceptions of wildlife, preferences for management, use of public lands and protected areas, or attitudes toward a proposed policy change.Social scientists develop indices from survey responses using various statistical techniques for factor analysis or clustering to aid in the analysis of social data (Gerbing & Anderson, 1988;Reise et al., 2000).Typically linked to geospatial information, these social indices can be mapped to visualize and analyze how social dimensions vary with environmental variables or cultural and political boundaries.Such mapping elucidates variation that cannot be uncovered with statistical analysis, and assists with the "prediction, understanding, and management" of social conflicts over conservation (Lecuyer et al., 2022, p. 1).For instance, social acceptance of carnivore populations and management in Europe varied with landcover, development, place-based traditions like shepherding or hunting, and local histories with wildlife (Gangaas et al., 2015;Piédallu et al., 2016).In Nepal, negative attitudes toward tigers spatially correlated with low socioeconomic status, fewer educational opportunities, and increased sociocultural marginalization, rather than with the distribution of livestock attacks (Carter et al., 2014).In the U.S., Manfredo et al. (2021) mapped wildlife value orientations and found that their distribution predicted state-level support for environmental policies like species reintroduction.Although addressing social conflict remains crucial for good wildlife governance (Dickman, 2010;Pomeranz et al., 2021;Redpath et al., 2013), few geospatial studies have demonstrated how social-ecological variation can address public conflicts over wildlife and inform integrated management strategies (Bergsten et al., 2014;Manfredo et al., 2021).
Social landscapes that include people's perceptions, attitudes, behaviors, and political or economic dynamics can productively be integrated with biophysical landscape and ecological factors (e.g., wildlife population densities, distributions, and migrations) to develop models of anthropogenic resistance (Ghoddousi et al., 2021), 'spatial coexistence' (Carter et al., 2020), or 'conservation conflict hotspots' (Lecuyer et al., 2022).Spatial modeling allows wildlife managers to identify where synergies or mismatches exist among wildlife population dynamics, habitat conditions, social tolerance or feasibility, and policy approaches, and then target local communities for management interventions (Dressel et al., 2018).Disregarding spatial diversity and local contexts, on the other hand, can "lead even highly advanced management approaches…into panacea traps" (Dressel et al., 2018:109;Zimmermann et al., 2021).Conservation programs that attend to local complexities (e.g., in cultures, livelihoods, institutions, and political-legal regimes) tend to find greater success in their social and ecological outcomes than those that focus on national contexts or do not consider local needs (Brooks et al., 2012;Galvin et al., 2018;Geoghegan & Renard, 2002;Hazzah et al., 2014;Waylen et al., 2010).Because human perceptions of wildlife populations-and their impacts or benefits-neither evenly nor predictably distribute across social and biophysical landscapes, neglecting to analyze this socio-spatial variation can hinder public support for conservation and management actions, reducing their efficacy (Carter et al., 2020).Since perceptions of wildlife and the acceptability of management covary with a suite of individual, social, and geographical variables (Lecuyer et al., 2022), social and spatial analyses should thus be combined to better understand how wildlife management can be tailored to local socialecological conditions (Dressel et al., 2018).

| STUDY CONTEXT AND OBJECTIVES
This study was situated in a North American context where state or provincial agencies are entrusted to manage wildlife populations for the potential benefit of all citizens, a framework known as the Public Trust Doctrine (Decker et al., 2016;Hare et al., 2017).Given the proliferation of large ungulates across North America, continued urbanization, and changing public values for wildlife (Campa III et al., 2011;Dietsch et al., 2016;Manfredo et al., 2009), a substantial body of research has examined human-ungulate conflicts and human-human conflicts over ungulate management strategies.This literature provides important insights into factors that shape such conflicts including wildlife value orientations, narratives and problem discourses, affective and emotional costs, conflicting management objectives, social identities, and political processes (e.g., Epstein & Haggerty, 2022;Haggerty et al., 2018;Hill, 2015;Knackmuhs et al., 2019;Leong, 2009;Leong et al., 2007;Peterson et al., 2013;Vernon & Clark, 2016;Zimmermann et al., 2020).In the case of white-tailed deer (Odocoileus virginianus, hereafter "deer"), intergroup conflict often arises from different wildlife values and concerns about deer-related impacts (Campa III et al., 2011;Connelly et al., 1987;Johnson & Horowitz, 2014;Lischka et al., 2008).Although hunters tend to harbor concerns about deer population sizes and hunting opportunities, they can also disagree among themselves about hunter recruitment, weapon use, quota distributions, and culling or sharpshooting methods (Stewart, 2011).While agricultural producers and woodland owners are typically concerned about damage to their lands, urban residents carry substantial concern for both public health and deer welfare (Stinchcomb et al., 2022a).
So far, much of the literature on human interactions with ungulates, and particularly deer, has been qualitative or descriptive in nature.Some studies have used various social-psychological or governance frameworks to better integrate social, cultural, economic, political, and institutional factors in understanding human-ungulate interactions (e.g., Heberlein, 2002;Loker et al., 1999;Peterson et al., 2002;Urbanek et al., 2012).In parallel, researchers have used population ecology or harvest management lenses to predict ungulate distributions and illuminate the ecological dimensions of human-ungulate conflicts (e.g., Hegel et al., 2009).However, few studies have analyzed the potential for social conflict over ungulate management approaches in a spatially explicit manner.
Specifically in our study context of Indiana, USA, the social landscape around deer remained largely unquantified until recently.Empirical social research has since found that Indiana residents hold complex values, attitudes, emotions, and beliefs about white-tailed deer that do not sort nicely into specific stakeholder categories (Stinchcomb et al., 2022a).Rather, people's perceptions of deer and deer management often conflict with one another, both within and among social categories (e.g., self-identity, gender, etc.).When the acceptability of different management strategies varies, the potential for policy or management-related backlash increases.Here, we seek to understand how social acceptability of deer management options in Indiana varies across self-identity categories and geographic gradients.We quantify potentials for conflict over deer management methods among residents of Indiana and translate them into hotspots of social conflict across the state.In so doing, we provide managers with a feasible, empirical approach to identify among whom and where high public conflict may occur.This approach can help managers develop new strategies to manage the social landscape in concert with wildlife populations and improve the overall social acceptance of deer management in Indiana and human-wildlife interactions more broadly.

| Data collection
This study analyzed the same large dataset from a statewide survey described in Stinchcomb et al. (2022b) and implemented from June to September 2021.Survey questions aimed to quantify the perceptions of Indiana residents related to deer populations, deer management, and the state's managing agency: the Indiana Department of Natural Resources (IN-DNR) Division of Fish & Wildlife (DFW).Details of the survey design, sampling, and dissemination can be found in Stinchcomb et al. 2022b.We include a copy of the survey questionnaire in Data S1 for reference to the structures of our survey questions and responses.Stinchcomb et al. (2022b) used the survey data to develop a multidimensional construct of public satisfaction with deer management and assess the predictors thereof.Building upon that work, this article examines how social conflicts over deer management methods are distributed within and among social groups and spatially across the state.
In this article, we focused on survey questions that measured how respondents evaluated the acceptability of six potential deer management methods in their areas: (i) increased licensed hunting; (ii) culling or sharpshooting; (iii) Community Hunting Access Programs (CHAPs); (iv) contraception; (v) translocation; and (vi) providing information.We measured each management method on a 5-point scale including a neutral value from "very unacceptable" (À2) to "neither" (0) to "very acceptable" (2).The IN-DNR currently uses licensed hunting, culling or sharpshooting, CHAPs, and providing information to manage deer in Indiana.Contraception and translocation options are not currently used, but they have been tested on deer in Indiana and elsewhere and are commonly discussed as non-lethal methods for managing ungulate populations (Demarais et al., 2012;DeNicola et al., 1996DeNicola et al., , 1997aDeNicola et al., , 1997b;;Rutberg et al., 2004;Sanborn et al., 1994;Swihart & DeNicola, 1995;Walter et al., 2010).

| Potential for conflict indices
The Potential for Conflict Index (Manfredo et al., 2003;Vaske et al., 2010) has been widely applied to humanwildlife interactions to understand how how conflicts over conservation or management strategies distribute among social groups (e.g., Liordos et al., 2017;van Eeden et al., 2019).The PCI also visualizes social conflict levels among stakeholders or demographic groups, value orientations, attitudes, or other social metrics (Vaske et al., 2006).
We analyzed potentials for social conflict over deer management methods in Indiana using the Potential for Conflict Index II (PCI 2 ; Vaske et al., 2006).We compared mean acceptability and PCI 2 values across the six deer management methods for two demographic groups of interest: respondents' primary self-identity and their political ideology.Self-identities were derived from the question: "What is your primary identity?Check one that you identify with the most:" (i) farmer or grower; (ii) rancher or livestock producer; (iii) woodland owner; (iv) deer hunter; (v) urban area resident; and (vi) rural area resident.Since we had only 42 responses (2%) in the rancher category, we combined farmers/growers and rancher/livestock producers into one "farmer/rancher" category for analysis.A multiple-identity question preceded this (Data S1) in which respondents could select every identity with which they identify.We choose to analyze the "primary identity" question to streamline our statistical analysis and use categories that are readily applied by deer managers in North America.Although deer hunters live in both urban and rural areas, we expected respondents with an active hunting license to identify primarily as deer hunters and non-hunting residents to identify with the other categories.To elicit political ideology, we asked respondents "Which of the following best describes your views?" on a 7-point scale: (i) strongly liberal; (ii) liberal; (iii) slightly liberal; (iv) middle-of-the-road; (v) slightly conservative; (vi) conservative; and (vii) strongly conservative.For analysis, we combined strongly liberal and liberal into a "Liberal" category, middle-of-the-road and slightly conservative into "Moderate;" and conservative and highly conservative into "Conservative."The slightly liberal option had no responses.
We used one-way analysis of variance (ANOVA) to compare the mean acceptability of each management method among respondent self-identities and political ideologies (Engel et al., 2017;Sponarski et al., 2015).Although scholars and statisticians debate the application of ANOVA to Likert-scale ordinal data (e.g., Carifio & Perla, 2008;Jamieson, 2004), likert scales can be and frequently are treated as intervals in parametric analyses without invalidating conclusions (Norman, 2010).
Following Sponarski et al. (2015), we used Bonferroni post hoc tests to determine significant differences in means among groups at adjusted p-values of 0.005 for self-identity (10 comparisons) and 0.0167 for political ideology (3 comparisons).When our data violated the homogeneity of variance assumption, we used Welch's ANOVA with a Games-Howell post-hoc test for group differences.This pairwise testing method is robust to differences in group sample sizes and error variances (Games & Howell, 1976;Stoline, 2012).We used R v.4.1.2base and rstatix packages for all ANOVA testing.
We calculated PCI 2 values within self-identity and political ideology groups for each management method using a Microsoft Excel spreadsheet available at https:// sites.warnercnr.colostate.edu/jerryv/potential-conflict-index/(Vaske et al., 2006).The PCI 2 calculates the ratio of responses on either side of a scale's center point so that maximum conflict (PCI 2 = 1) occurs when responses are evenly split between the two extreme values and total consensus (PCI 2 = 0) occurs when all responses fall on a single value in the scale (Engel et al., 2017;Liordos et al., 2017;Vaske, 2018).We used a distance function that excludes the neutral value from PCI 2 calculations and a power of 1.5 to allow for nonlinear perceptions of the differences between scale values.A nonlinear power function does not conflict with the linearity assumption of ANOVA because ANOVA only compares mean acceptability levels among groups, not PCI 2 values within groups.
We tested for differences in PCI 2 values between pairs of groups using the difference test provided on the PCI 2 website and compared the resulting d value to critical values of a standard normal distribution, at an alpha level of 0.05 (Engel et al., 2017;Sponarski et al., 2015).If d > 1.96, we deemed the difference between PCI 2 values significant (Vaske et al., 2010).Detailed equations for PCI 2 calculations and tests are provided in Supporting Information.

| Mapping social conflict potentials
Using ArcGIS Pro (ESRI Inc. 2021), we created heat maps of social conflict pertaining to deer management across Indiana using respondent acceptability ratings for our six deer management methods: culling, licensed hunting, CHAPs, contraception, translocation, and providing advice or information.Since we aimed to map areas where the difference between respondents' acceptability values was highest, a conventional hotspot analysis-which determines areas where high values are surrounded by other high values or low values by other low values-did not work for our purpose.Following Brown et al. (2017) and Moore et al. (2017), we overlayed a 12.9 Â 12.9 km (8 Â 8 mi) grid over our final survey sample derived from a 6.4 Â 6.4 km (4 Â 4 mi) statewide grid created by the IN-DNR and used by our colleagues for ecological sampling (Caudell & Vaught, 2018).Using the same grid allows the IN-DNR to easily integrate our spatially explicit social data into existing ecological analyses like the generation of Regional Management Units for deer.After testing several grid sizes (see Supporting Information), we chose a 12.9 km 2 (8 mi 2 ) cell area to balance the number of points within each grid cell with the spatial resolution required to assess local, place-based trends in conflict potentials (Brown et al., 2017).Due to the dispersal of our survey points in the most rural areas of the state, a smaller grid size (6.4 km 2 /4 mi 2 ) would result in too many zero values in the conflict analysis, while a slightly larger grid size (19.3 km 2 /12 mi 2 ) was too regional, masking local variability particularly around urban centers (see Supporting Information).
Using survey responses from the sample points within each grid cell, we calculated the mean and frequency distribution of acceptability responses and the PCI 2 value for each management method.We then used the gridded PCI 2 values in the Hotspot Analysis (Getis-Ord Gi*) Spatial Statistics tool (ESRI, 2021) to visualize where grid cells with high potentials for social conflict were surrounded by other grids with high social conflict potentials and, by opposition, areas where social conflict was low.Unlike other interpolation methods such as kernel density estimation, the Gi* statistic method required little prior knowledge about clustering and measured the statistical significance of the hotspots generated.Grid cells with no sample points or only one sample point were excluded from the hotspot analysis to avoid zero inflation.We created one heat map for each of the six management methods.
We chose a fixed distance band method for the hotspot analysis to analyze the spatial relationships among gridded PCI 2 values.We used the Incremental Spatial Autocorrelation analysis to determine the optimal search distance for each management method.The Hotspot Analysis tool calculates the Getis-Ord Gi* statistic as a z-score (Getis & Ord, 1992).Large, positive z-scores with small p-values indicate significant clustering of high PCI 2 values and produce a hotspot on the map.Large, negative z-scores with small p-values indicate significant clustering of low PCI 2 values and produce coldspots.Based on the distances at which z-scores peaked for each method, and opting for more regional rather than local clustering, we used a distance band of 32 miles (51.5 km).Significant hotspots and coldspots were mapped with color gradients to indicate the level of confidence in the statistic: 90%, 95%, or 99% confidence.We overlaid our final heat maps with the boundaries of the IN-DNR's Regional Management Units (RMUs) for deer to illustrate how social conflict could impact Indiana's deer management strategy.T A B L E 1 Summary statistics for respondent acceptability ratings of six potential deer management methods in Indiana (N = 1806).

Variable name
Variable description and value levels n

Proportion of responses
Primary self-identity a

| Descriptive statistics
We obtained a 33% response rate on the survey (1806 complete responses out of 5500 eligible and delivered).
Responses were relatively well distributed across the state, with some clustering evident near major cities (Figure 1).When asked to select the group with which they identify the most, 43% of respondents identified as primarily rural residents, 25% as primarily urban residents, 12% identified as primarily farmers or ranchers, 12% as primarily deer hunters, and 8% as primarily woodland owners (Table 1).Among licensed hunters who responded to the identity question (n = 463), 80% identified as primarily deer hunters, 20% as primarily rural residents, and 10% as primarily urban residents.
Most respondents reported a conservative political ideology (59%), with 22% aligning with a moderate ideology and just 12% aligning with a liberal ideology (Table 1).
A detailed summary of survey respondents' demographics can be found in Stinchcomb et al. (2022b).In terms of respondents' perceptions of deer management, just under half of respondents found licensed hunting (49%) and information provision (47%) to be acceptable management methods (Table 1).Respondents were split on the acceptability of culling and CHAP methods, with 35% rating culling as unacceptable and 34% as acceptable, and 33% rating CHAPs as unacceptable and 32% as acceptable (Table 1).Contraception and translocation were the least acceptable management methods among respondents, with 43% rating contraception as very unacceptable and 38% rating translocation as very unacceptable (Table 1).

| Potentials for conflict among Self-Identified stakeholder groups
Self-identified stakeholder groups showed differences in mean acceptability ratings across nearly all deer management methods, except for providing information which was acceptable for all groups (means = 0.24-0.58; Figure 2).The greatest differences occurred for nonlethal methods of contraception and translocation.Deer hunters rated these nonlethal methods as significantly less acceptable than all other groups (x cont = À1.66;F 4,537 = 73.70,p allÀcontrasts < 0.001; x trans = À1.33;F 4,524 = 36.62,p allÀcontrasts < 0.001).Urban residents rated contraception and translocation as significantly more acceptable than all other groups, but their average rating was still slightly unacceptable (x cont = À0.27;F 4,537 = 73.70,p allÀcontrasts < 0.001; x trans = À0.16;F 4,524 = 36.62,p allÀcontrasts < 0.001).Farmers, woodland owners, and rural residents all rated nonlethal methods as unacceptable on average (Figure 2).Mean acceptability ratings for lethal methods also differed between self-identity groups.Farmers showed significantly greater acceptability of deer culling, on average (x farm = 0.26), than either deer hunters (x hunt = À 0.57) or woodland owners (x wlo = À0.28;F 4,509 = 12.06, p F,H < 0.001, p F,W = 0.005).Deer hunters also rated culling as significantly less acceptable than did rural and urban residents (x hunt = À 0.57; x urban = À 0.02; x rural = 0.05; F 4,509 ¼ 12:06, p H,U , p H,R < 0.001).Licensed hunting was the most acceptable management method across self-identity groups (Figure 2).Mean levels of acceptability differed significantly, however, between farmers and hunters, with farmers rating hunting as more acceptable than hunters themselves (x farm = 0.67; x hunt = 0.18; F 4,1713 ¼ 5:63, p F,H = 0.001).Acceptability of CHAP programs was neutral, on average (Figure 2), but woodland owners showed significantly lower acceptability of CHAPs than urban residents (x wlo = À0.33;x urb = 0.03; F 4,1684 = 2.42, p W ,U = 0.032).Levels of inter-group conflict were thus highest for culling, contraception, and translocation methods (Figure 2).Licensed hunting also showed potential conflict between farmers and hunters.
Potentials for conflict within each self-identity group were highest for hunters over increases in licensed hunting (PCI 2 = 0.48), and lowest for hunters over contraception methods (PCI 2 = 0.09).Farmers also showed relatively high intragroup conflict over culling, hunting, and contraception methods (PCI 2 range 0.36-0.40).Compared to farmers, intragroup conflicts over culling and CHAPs were significantly lower for urban residents (Figure 2).Intragroup conflict over contraception was significantly lower for hunters than for all other groups (Figure 2).Hunters also showed significantly higher intragroup conflict over licensed hunting than all groups except woodland owners and significantly lower intragroup conflict over translocation than urban and rural residents (Figure 2).
Although levels of intragroup conflict did not differ significantly across political ideologies for culling and licensed hunting methods, liberal respondents showed the highest PCI 2 over these methods (each PCI 2 = 0.40).Their intragroup conflict over contraception was significantly higher than that of both other ideologies and intragroup conflict over translocation was significantly higher than that of conservative respondents (Figure 3).Respondents who identified with a moderate ideology showed significantly lower intragroup conflict over CHAPs than both conservative and liberal respondents.For all groups, the lowest PCI 2 occurred for providing information about deer management (PCI 2 range = 0.20-0.21).

| Hotspots of social conflict over deer management
We located the largest and most significant hotspots of social conflict over deer management for culling, licensed hunting, and contraception methods.The largest hotspot for culling appeared in southern Indiana, with 87% concentrated within a single RMU (Figure 4a).Another significant culling hotspot with 90%-95% confidence occurred in northeast Indiana, crossing the border of two RMUs (Figure 4a).Conflict over licensed hunting was also significant in this area, but only a few grid cells showed high conflict potentials in southern Indiana.The largest hotspot for hunting appeared in central Indiana, concentrated in the Indianapolis metropolitan area, which comprises its own RMU, but spilling over into two RMUs to the east (Figure 4b).Following from the findings above, social conflict over CHAPs was relatively low.A single hotspot of six grid cells with 90-95% confidence emerged in northeast Indiana (Figure 4c), overlapping with a section of the corresponding hotspots for licensed hunting and culling.
For all lethal methods, significant coldspots were found in three main regions.These occurred in the westcentral area and the southwest tip of the state, with a few cells in the southeast showing low social conflict with 90% confidence (Figure 4a-c).Cold spots with 95% confidence appeared in the southwest for culling, the west-central for hunting, and the east-central for CHAPs.Most coldspots fell within a single RMU, except for one or two grid cells of the west-central hotspots crossing RMU borders for both culling and hunting methods.
Among nonlethal methods, the largest and most significant hotspot of social conflict appeared for contraception in the southeast corner of the state (Figure 4d).This hotspot spanned almost the entirety of a small RMU and crossed into its western neighbor.Smaller hotspots appeared for both contraception and translocation methods in the northern region of the Indianapolis metropolitan area in central Indiana (Figure 4d, e).Additional hotspots occurred for translocation near the southern border of the state and, for contraception, in the south-central region.The information provision method showed a few areas of significant social conflict, mainly in south-central and northwestern Indiana (Figure 4f).
Areas of low social conflict differed across nonlethal management methods.For contraception, a few cold spots occurred in northern rural areas, whereas a more concentrated cold spot with 90%-95% confidence appeared for translocation in the west-central part of the state (Figure 4d, e).Just four grid cells showed cold spots with 90%-95% confidence for information provision, all of which appeared in central Indiana (Figure 4f).These areas of low social conflict over nonlethal methods were mostly within a single RMU, except for one or two cold spot cells that crossed an RMU border for each management method.

| DISCUSSION
Our study combined statistical and geospatial tools to assess where and among whom social conflicts over wildlife management methods are occurring or could occur in the context of white-tailed deer.Previous scholarship demonstrates that stakeholder identities explain differences in the acceptability of wildlife management methods (Bruskotter et al., 2009;Liordos et al., 2017;Lohr & Lepczyk, 2014;Messmer et al., 1997;van Eeden et al., 2019).Our findings confirm that acceptability of deer management methods varies with resident selfidentities, but the pattern of certain conflicts may be unexpected by wildlife managers.Specifically, we have two unexpected results.First, deer hunters showed the highest level of intragroup conflict over licensed hunting among all identity groups.This counterintuitive result likely stems from mixed beliefs among hunters about increased hunting activity and its effects on crowding, availability of game, and access to private hunting lands (Heberlein, 2002;Larson et al., 2014).This result highlights the heterogeneity in environmental and natural resource perceptions and behaviors within groups that have been traditionally thought of as homogenous.Second, although urban residents did show greater acceptability of nonlethal methods than any other identity group, similar to what has been found in the literature (Messmer et al., 1997;Stewart, 2011;Urbanek et al., 2012;Wilkins et al., 2019), self-identified rural and urban residents in our study did not differ in their acceptability of lethal deer management methods (Figure 2), different from the literature describing urban-to-rural divides over hunting (Ericsson et al., 2018;Heberlein & Ericsson, 2005;Wilkins et al., 2019).Although we recognize that "rural" and "urban" categories are not monolithic, this result highlights how groups typically perceived as conflicting can share environmental attitudes or values within a specific context.
Together, these two unexpected results are illuminating because of the constant need to balance the efficiency and efficacy of conservation programs.On one hand, conservation programs are expected to be generally applicable to a wide diversity of people and situations.On the other hand, there is an increasing recognition of the importance of tailoring conservation program design and implementation to incorporate social, cultural, and economic diversities to address diverse community needs and concerns.In a way, researchers and conservation practitioners have been using different approaches to sort stakeholders and the public into "bins" based on their sociodemographic and other observable characteristics and working to make conservation programs applicable and appealing to each "bin" of people.The tradeoff remains that too many "bins" are unworkable but too few "bins" are ineffective.Our results shed light on this tradeoff in the context of wildlife management and conservation and highlight the need for more research to inform the determination of the number and characteristics of these "bins" to balance efficiency and efficacy in conservation programs.Our results also provide further evidence that individual self-identities may differ from how wildlife managers typically delineate stakeholder groups (Stinchcomb et al., 2022a).Thus, wildlife practitioners should not automatically assume preconceived stakeholder categories as a reliable predictor of the social acceptability of wildlife policies and programs.It is also important to keep in mind that one's self-identity can vary with the object in consideration like a wildlife species versus a policy goal (Lute & Gore, 2014), and that individuals may identify with multiple, intersecting social groups (McCubbin & Van Patter, 2021).Our analyses focused on the self-reported primary identity of survey respondents, but we acknowledge that a more nuanced analysis of social identities using Likert-type scales of affiliation should be increasingly used in survey research within the human dimensions of wildlife field (Grooms et al., 2022;Schroeder et al., 2021;Snyder et al., 2022).
In addition to these insights about self-identity, our study also highlights the role of political ideology in shaping human-wildlife interactions and the associated social conflicts.Previous research has suggested that political ideologies can influence trust in wildlife agencies (Manfredo et al., 2017;Schroeder et al., 2021) and interact with wildlife values to influence risk perceptions and attitudes toward wildlife (Nardi et al., 2020).We found that political ideologies produced expected differences in the social acceptability of deer management.Residents with liberal ideologies reported opposing and significantly different mean acceptability levels from those with conservative ideologies across all management methods except culling.Political values tend to be stable over time, particularly at ideological extremes (Goren, 2005;Zaller, 1991).Compared to selfidentity, political ideology may be a more consistent predictor of the acceptability of wildlife management.When examining social acceptance or tolerance, particularly in the North American context, wildlife managers should consider residents' political ideologies and how these influence the processing of information about the impacts and benefits of wildlife populations (Hart & Nisbet, 2012).
Yet political ideology is likely not the sole driver of different perceptions of wildlife or management acceptability.Rather, political ideology reflects long-term trends in the United States and elsewhere of geographic clustering of self-similar social groups (Bishop & Cushing, 2009).Liberal affiliations tend to concentrate in larger urban areas while smaller cities are more politically mixed, and suburban fringes become increasingly conservative (Gimpel et al., 2020).Our sample of self-identified urban residents were twice as liberal (22%) as rural residents (12.5%), less conservative (54%) than rural residents (64%), but with similar proportions identifying as politically moderate (23.7% moderate/urban, 23.6% moderate/rural).We found that liberal-leaning (likely urban) residents were generally less accepting of licensed hunting and CHAPs than other political ideologies, potentially leading to higher conflict with moderate and conservative urbanites.For example, residents of Fort Wayne, where social conflict hotspots occurred across lethal methods, tended to self-identify as politically moderate and conservative with a small patch of liberal precincts near the city center (Dottle, 2019).Although Fort Wayne is inside an Urban Deer Reduction Zone (DRZ), the diversity of hunting attitudes among its residents produces mixed acceptance of increased hunting activity (Stewart, 2011).Whereas many rural landowners participate in hunting or engage directly with hunters, proportionately fewer urban residents participate in hunting despite their proximity to special hunting opportunities like DRZs or CHAPs (Stewart, 2011;Wilkins et al., 2019).Additionally, our largest and most significant hotspot of social conflict occurred over culling around Bloomington, Indiana, a metropolitan region classified as suburban and politically mixed and where past controversies over deer culls have occurred (Knackmuhs et al., 2019;Knackmuhs & Farmer, 2017).Hunting in urban areas is logistically challenging, and urban residents tend to have less experience with and perceive greater safety risks from hunting than rural residents (Heberlein & Ericsson, 2005;Kilpatrick et al., 2007).While in the aggregate, self-identified urban residents do not differ from rural residents in their acceptability and conflict over lethal deer management, variations in PCI values can scale with the size and sprawl of the urban area in question.For this reason, political affiliations better align with the spatial distribution of social conflicts than other sociodemographic categories.High social conflicts in and around cities are likely driven by the interaction between differential experiences with wildlife management, conflicting political ideologies, and contrasting values for wildlife.We encourage researchers and practitioners to further examine the diversity within the "urban resident" category to parse out differences in experiences, values, and management preferences among residents of exurban, suburban, and urban zones and how such preferences align with ideological gradients.
Social conflict analyses, like the PCI, can help practitioners identify intra-and inter-group conflicts that would (or do) challenge conservation approaches and begin to address them through targeted engagement.For example, disagreement between hunters and farmers complicates wildlife management and conservation on private lands, where agencies rely on landowners to provide access to hunters both to control wildlife populations and address issues like crop depredation (Kamal & Grodzinska-Jurczak, 2014;Sorice et al., 2011).Although the farmers in our study reported general acceptance of increased hunting licenses, hunters still face pushback from farmers or other private landowners when requesting access to private lands due to landowner concerns about trespassing, safety, or hunting ethics (Stinchcomb et al., 2022a).We suggest that agencies and organizations try to disentangle the complexity of identity-perception relationships when they have the resources to extensively survey their publics.If conflicts persist, practitioners should facilitate collaborative discussions with conflicting and counterintuitive groups like different types of hunters or urban and rural residents to understand the drivers of acceptability and conflict and attempt to arrive at shared goals for management or conservation (Lute & Gore, 2014;Redpath et al., 2013).To effectively find common ground, these discussions should span multiple issues rather than focusing on a single conflict (Lecuyer et al., 2018).Where and when such resources run thin, agencies and organizations may need to rely on more stable, predictable measures like voter behavior, political ideology, or value orientations, which can be accessed from other sources such as social research or polling organizations.
Using social conflict analysis, practitioners can also determine with whom and where they need to improve information transparency and diversify their toolkits to increase public support for conservation.Public support, or a lack of public opposition, is critical for effective policy implementation (Sanborn et al., 1994), but the public often lacks awareness about the cost, implementation, and efficacy of various wildlife management methods (Kilpatrick et al., 2007;Walter et al., 2010).While some management strategies, particularly the provision of information, were viewed as acceptable across different social categories, they are neither cost-nor risk-free; as the COVID-19 pandemic showed, information provision can quickly become politicized or subject to misinformation campaigns.Nonlethal methods like contraception tend to face conflicting public perceptions about their humaneness and their effects on hunting traditions (Curtis et al., 1993;Kilpatrick & Walter, 1997).Even lethal methods face challenges of changing public values for wildlife and declining hunter populations (Jacobson et al., 2010;Manfredo et al., 2009;Price Tack et al., 2018).Wildlife agencies in the U.S. start to recognize the need to expand funding for wildlife management and conservation beyond license revenues and equipment taxes, but this requires a continuous shift in wildlife agency values and in the 'beneficiaries' that wildlife managers consider when making decisions (AFWA & WMI, 2019;Decker et al., 2016;Hare et al., 2017;Jacobson et al., 2010;Serfass et al., 2018).For example, urban residents are an emerging target for broadening public support for wildlife and natural resource management both in the U.S. and globally (Davies et al., 2004;Egerer & Buchholz, 2021;McCance et al., 2017).But, as our study highlights, urbanites hold multiple different management preferences, and those preferences will likely be more diverse when broken into sub-groups (e.g., urban core, suburban, and exurban; hunting and nonhunting).Social conflict potentials add to the challenge of managing humans and wildlife in urban areas, due to diverse demographics, municipal authority, varying land use or zoning, lower hunting opportunities, and greater concerns about hunting risks.Along with public engagement about lethal management methods, nonlethal methods should be included in management strategies for urban areas because both social and ecological feasibility are required for effective wildlife management (Brown et al., 2019;Clifford et al., 2022;Lischka et al., 2018).Depending on wildlife agency and organization funding and the dynamics of local wildlife populations, urban areas could serve as test grounds for mixed lethal and nonlethal management strategies.
Mapping social acceptability and conflict over space has broad implications for meaningfully integrating social information into conservation planning from local to regional scales (Brown et al., 2019;Moore et al., 2017).There is an increasing recognition of the importance of "integrating social factors into traditional spatial analysis to promote human-wildlife coexistence" (Carter et al., 2020, p. 1;Lecuyer et al., 2022).Empirically, Carter et al. (2014) mapped the attitudes of local community members toward tiger in Nepal to inform tiger conservation planning and engagement efforts in and around a national park.More recently, researchers developed a spatially explicit model to incorporate anthropogenic and environmental variables in predicting human-caused brown bear mortality in Iran (Nayeri et al., 2022).So far, many of these analyses have been conducted on large, charismatic, often carnivorous species, and fewer studies have analyzed the social conflicts associated with non-carnivorous species, like ungulates, in a spatially explicit manner.For example, in Indiana, RMUs for deer were delineated based on models of deer mortality, habitat characteristics, and landcover including levels of urban development (Swihart et al., 2020).These units do not capture social variation over space, because most wildlife management strategies remain driven by the ecology of the species and landscape.Conflicting or negative social perceptions of wildlife management, however, can create unique challenges like public opposition toward agencies, voter-mandated management actions, and exacerbated human-wildlife conflict (Manfredo et al., 2017;Minnis, 1998;Redpath et al., 2013;Williamson, 1998).If the social dimensions remain unaddressed by wildlife agencies and organizations, social conflicts will increase the politicization of wildlife management and conservation (Ditmer et al., 2022) and decrease the efficacy of related programs, especially those aiming to reduce human-wildlife conflicts (Bhatia et al., 2020;Dickman, 2010).
Our study also highlights the potential for conservation practitioners to use spatial information about social conflicts to delineate management units that are both ecologically and socially suitable for wildlife species or conservation interventions (Dando et al., 2023;Ditmer et al., 2022;Sage et al., 2022).Our hotspot analysis highlighted areas of social consensus (coldspots) where management methods will receive little public opposition and areas of social conflict (hotspots) where management will likely be met with public backlash.This approach can be used more broadly by wildlife agencies and organizations to target areas for public engagement around a specific conservation or management strategy.Our study also provides clear documentation of how to conduct this kind of hotspot and coldspot analysis, which can be applied to other species, strategies, and interventions in other locations.Beyond engagement, wildlife agencies and organizations can follow methods employed for delineating wildlife management units (e.g., RMUs, Swihart et al., 2020) or protected areas to create social management units that encompass regions where public conflict levels, attitudes, and ideologies are similar.Wildlife practitioners can then analyze the conditions within social and ecological units simultaneously to determine the most suitable conservation strategy for each region (Behr et al., 2017).If mismatches occur between social perceptions and ecological conditions, these areas could be targeted for social interventions or negotiations of management and conservation priorities (Behr et al., 2017;Bergsten et al., 2014;Dressel et al., 2018).

| FUTURE RESEARCH & STUDY LIMITATIONS
Future studies should advance our spatial analyses of social conflict over wildlife management and integrate this information with ecological variables.While other studies have mapped the influence of population and landscape variables on human-wildlife conflicts (Carter et al., 2014(Carter et al., , 2020;;Sharma et al., 2020;Struebig et al., 2018;Tripathy et al., 2021), the influence of these variables on social conflicts over wildlife has yet to be fully assessed.Since the persistence of social conflicts hinders effective wildlife governance (Dickman, 2010;Pomeranz et al., 2021), and value-based differences typically underlie those conflicts (Bruskotter et al., 2009;Schroeder et al., 2021), conflicts over the acceptability or tolerance of management strategies should be associated with spatial data on social values (political, ecological, and cultural) and correlated constructs like risk perceptions (Lecuyer et al., 2022).Still, tailoring wildlife management strategies to both social and ecological conditions remains challenging in practice.Effective social-ecological integration requires continued collaborations between wildlife biologists and social scientists, in spaces that promote reflexivity (Atkins, 2004), relational thinking (Cruikshank, 2005;Haraway, 1988;Latour, 2005), and negotiation of epistemological differences (Angelstam et al., 2013;Fielding, 2012).Interdisciplinary research also faces limited funding, training, leadership, and acceptance within natural resource agencies and even academic institutions (Jacobson et al., 2022;Teel et al., 2022).Although these institutional barriers persist, wildlife agencies and organizations can work to increase their social science capacities and shift their institutional cultures toward recognizing the importance of social data and local knowledge (Bélisle et al., 2018;Jacobson et al., 2022;Manfredo et al., 2019;Morales et al., 2021).Geospatial analysis provides one relatively accessible tool for wildlife agencies and organizations to begin integrating complex social and ecological landscapes and viewing the relationships between humans and wildlife through cross-scalar and cross-disciplinary lenses (Fielding, 2012;Teixeira, 2016).
Given the local to global relevance of our study, we want to recommend caution when extrapolating our results broadly.Compared to census data for Indiana, our sample contained significantly greater proportions of white/Caucasian, male, well-educated, high-income, and rural-dwelling residents (Table 1).These differences are directly attributable to our sampling strategy, which intentionally targeted rural forestland and agricultural properties, as well as existing customers of the Indiana DFW.We encourage future research to sample more representatively from urban-rural gradients and explore divides among these groups in greater depth.For instance, delineating sub-groups of the "urban resident" category (e.g., resident of exurban, suburban, or urban zones) could capture the social and spatial variation in perceptions within this commonly used category more effectively.Our analysis was further limited by the size of grid cells used for the hotspot analysis.Grid cells with only one data point were excluded from the hotspot analysis because of a zero PCI 2 value, which limited our ability to detect significant hotspots and coldspots in very rural areas of the state.Additionally, the intensity and distribution of hotspots can shift with the grid cell size chosen during analysis (Figures S1 and S2).
The selection of grid cell size depends substantially on the distribution of available social data, highlighting the challenge of mapping conflict using point-based data.Our use of the Gi* hotspot method worked well for points that were relatively dispersed throughout the study area and our interest in eliciting relatively local variability.When applying our methods to other species and in other locations, researchers and practitioners may benefit from expanding data collection efforts to achieve a more even distribution of survey responses across the geographic area of their interest.Future analyses of social conflict potentials over space may also benefit from testing different interpolation methods and their utility with point-based survey data.We recommend that practitioners choose both a grid size and hotspot analysis method that makes sense for their sample of social information and the scale at which they wish to make management or conservation decisions.

F
I G U R E 1 Distribution of survey responses across Indiana, overlayed on Regional Management Units for deer and incorporated areas of the 15 largest cities by population.IDMP = Integrated Deer Management Project.
Mean acceptability levels of six potential deer management options and Potentials for Conflict (PCI2) among respondents' primary self-identities.Bubble size represents PCI2 value, which is provided to the right of each bubble.Superscript letters (a,b,c) on PCI2 values represent significant differences in conflict potentials among the five groups.Numbers within bubbles (1, 2, 3) represent significant differences among mean group acceptability responses.If groups do not share a number, they have statistically different mean acceptability ratings (p < 0.05).Color indicates group affiliation.
Mean acceptability levels of six potential deer management options and Potentials for Conflict (PCI2) for liberal, moderate and conservative respondents.Bubble size represents PCI2 value, which is provided to the right of each bubble.Superscript letters (a,b,c) on PCI2 values represent significant differences in conflict potentials among the five groups.Numbers within bubbles (1, 2, 3) represent significant differences among mean group acceptability responses (p < 0.05).If groups do not share a number, they have statistically different mean acceptability ratings (p < 0.05).Color indicates group affiliation.F I G U R E 4 Hotspots of high social conflict and coldspots of low social conflict over deer management methods in Indiana, USA.Culling (a), licensed hunting (b), and Community Hunting Access Programs (c) are lethal methods.Contraception (d), translocation (e), and providing information (f) are nonlethal methods.Data are overlaid with survey response points and Regional Management Units for deer.