Inadequate funding and lack of local acceptance and support are among the primary reasons why conservation efforts fail (Muhumuza and Balkwill, 2013). The “user-pay” model underpinning the North American system of conservation and the resulting institutional structures that emerged in the early 20th century were an opportune solution to integrate stakeholder participation and funding for wildlife management. Market hunting, resource exploitation, and habitat loss led to population declines of many wildlife species in the 19th century, and early conservationists viewed regulated hunting and population management as a way to democratize wildlife protection (Organ et al., 2012). Hunters and sportspeople were recruited as integral participants in this system of habitat management and regulated harvest, becoming the primary source of financial support and active management assistance to state wildlife agencies (Prukop and Regan, 2005). The Federal Aid in Wildlife Restoration Act (better known as the “Pittman-Robertson Act,” after the law's Congressional sponsors) was passed in 1937 and levied an 11% excise tax on firearms and ammunition (and later amended to include archery equipment) with proceeds distributed to state fish and game management agencies (Williams, 2010). Combined with state hunting license sales, these became the primary mechanisms for generating the states' funds for habitat management and acquisition (Organ et al., 2012).

This paradigm of wildlife conservation and management in North America, led by experts trained in biological sciences and aided by hunters, was successful in achieving several wildlife restoration objectives. Many species that were perilously close to extinction were recovered through habitat management and regulated hunting, including white-tailed deer (Odocoileus virginianus), wild turkey (Meleagris gallopavo), Rocky Mountain elk (Cervus elaphus nelsoni), and American beaver (Castor canadensis), among others (Organ et al., 2012; Heffelfinger et al., 2013). Some scholars, however, point to continued declines in biodiversity, habitat loss, and political barriers to maintaining support for wildlife management and conservation as evidence of needed reforms to address modern challenges (Feldpausch-Parker et al., 2017). Moreover, the historical reliance on hunters privileges certain subsets of wildlife-associated interests, which runs counter to the stated ideals of the public trust doctrine in managing resources for public benefit (Jacobson et al., 2010; Feldpausch-Parker et al., 2017).

Excise taxes derived from the Pittman-Robertson Act, state license sales, and general appropriations and other contributions provide between 60 and 90 percent of most state fish and wildlife agency budgets (Organ et al., 2012). However, state led conservation programs tend to focus primarily on game species while non-game conservation programs are often under-funded (Anderson and Loomis, 2006; Dalrymple et al., 2012). Several attempts have been made in the past at state and federal levels to diversify the portfolio of conservation funding sources and broaden user-based taxes and fees to include non-game species and incorporate other wildlife-associated recreationists (e.g., the “Teaming with Wildlife” initiative would have codified an excise tax on general outdoor gear). Many of these efforts have faced opposition from the outdoor industry and special interest groups (Secunda, 1998; Outdoor Industry Association, 2017). Additionally, reallocation of state sales taxes, lottery proceeds, and alcohol taxes, state wildlife grants, various Farm Bill programs, and various federal programs (e.g., Land and Water Conservation Fund) have thus far have not achieved the level of financial support needed to sustain wildlife conservation efforts (Mangun and Shaw, 1984; Franklin and Reis, 1996; Anderson and Loomis, 2006; Echols et al., 2019; Duda et al., 2021).

Future conservation successes may depend on expansion of sources for social and financial support, integration of non-game programs, and incorporation of broader societal interests and values toward wildlife into decision making processes (Nie, 2004; Jacobson et al., 2010; Decker et al., 2016). A better understanding of the factors that influence stakeholder support for alternative mechanisms for funding conservation may improve trust and transparency in decision making (Schroeder et al., 2021), leading to more broadly acceptable management decisions and enabling more effective governance of wildlife resources. Increased support for conservation funding may also help conservation advocates and decision makers build collaborative partnerships that alleviate political conflict and work toward sustainable funding models that incorporate the best interests of wildlife species and the diverse publics who value them.

As the diversity and popularity of non-hunting recreation has grown in North America, responsibilities for management have diversified from providing huntable game to providing for a range of wildlife tourism and recreation opportunities involving non-game species (Duffus and Dearden, 1990). Prior inquiries have evaluated support for conservation funding proposals that benefit both game and non-game wildlife species (e.g., Mangun and Shaw, 1984) and found support for a combination of funding sources, including preferences for excise taxes on general outdoor gear (Loomis and Mangun, 1987), sales taxes on outdoor gear (Dalrymple et al., 2012), and industry contributions (Larson et al., 2021). Socio-demographic characteristics often reflect cultural identities across rural-urban gradients, differing ethnicities, age groups, and gender identities that can influence views on wildlife conservation (Dalrymple et al., 2012; Kellert et al., 2017; Manfredo et al., 2017; van Eeden et al., 2020). However, achieving support for new laws or policies may also depend upon several social, political, economic, and individual factors. Evidence suggests that support for policies depends on the perceived burden to the individual (i.e., self-serving bias) (Caplan, 2011), whether the policy is voluntary or compulsory (Schroeder et al., 2020), as well as social-psychological variables such as personal values, social norms, and other contextual factors (Stern, 2000).

We designed a web-based survey to measure support for conservation policy proposals among a sample of Michigan hunters and wildlife watchers. In addition, we gathered information on relevant behavioral, social-psychological, and demographic variables we expected to be associated with support for conservation funding policies. The web-based survey was administered by the Michigan Department of Natural Resources' Marketing and Outreach Division in June-August 2020 as part of a broader research effort using a tailored design method for internet surveys (Dillman et al., 2014). The survey was presented on the Qualtrics platform and distributed to 522,993 addresses in the MDNR e-mail database. After removing out-of-state responses, incomplete surveys, and respondents who received the survey link through external sources, we obtained a final sample of 19,143 responses (4%).

Participation in nature-based activities was assessed to segment respondents into recreational typologies based on self-reported measures of participation. Nature-based activities in which each individual participated were ranked such that the most important recreation became their “primary recreational activity” and allowed us to assign each respondent a recreational type. Since only one activity could be ranked #1, respondents could be segmented according to their primary recreational activity. Those who chose either hunting (n = 2,558) or wildlife watching² (n = 942) as their primary recreational activity were included in the sample for the following analysis. Research protocols were approved by the Michigan State University Institutional Review Board (STUDY00001445).

Our dependent variables were measured by four items asking respondents to rate their intention to support hypothetical scenarios designed to increase funding for wildlife conservation efforts on a 5-point scale: strongly oppose (1), somewhat oppose (2), neither support nor oppose (3), somewhat support (4), or strongly support (5). Items were chosen based on a review of policies proposed or adopted in various states to provide broadened funding sources for wildlife conservation and were deemed to adequately reflect respondent attitudes toward regulatory, policy, or management solutions to conservation funding dilemmas. Four items were chosen for analysis to assess support for broad-based funding measures for wildlife conservation:

Items were recoded to create four binary outcome variables indicating “support” or “oppose” for each policy category. Responses were combined such that “somewhat support” and “strongly support” became one category coded as a (1) to indicate support, while “strongly oppose,” “somewhat oppose,” and “neutral” responses were coded as (0) to indicate responses that were not in support of the proposed policies. Neutral responses were assumed to align more with oppositional responses in that both categories reflect the absence of support for a given policy option.

Covariates hypothesized to be associated with support for conservation policies included membership in a stakeholder group (hunter or wildlife watcher), wildlife value orientations, cognitive behavioral antecedents (e.g., social norms, efficacy), natural and civic place attachment, past engagement in stewardship behaviors, recreation specialization, knowledge of conservation funding mechanisms, membership in a conservation organization, and socio-demographic characteristics. Wildlife value orientations reflect patterns of basic beliefs about the relationship between humans and wildlife. Generally, they fall along a spectrum from use-oriented (domination) perspectives and protection-oriented (mutualism) perspectives (Fulton et al., 1996). Confirmatory factor analysis supported the previously validated two-dimensional wildlife value orientation scales achieving high internal consistency (as measured by Cronbach's alpha) with domination (α = 0.73) and mutualism (α = 0.80) as two ends on a spectrum of wildlife beliefs (Manfredo et al., 2009). We used 13 items from scales adapted from previous studies of wildlife value orientations. Items were presented as belief statements from 1 (strongly disagree) to 5 (strongly agree). Items were averaged to create a continuous composite index on the original 1–5 scale for each wildlife value orientation.

Social norms refer to societal pressure for individuals to conform (or deviate) relative to others to commonly accepted modes of behavior, which guides individual actions. One item was presented as a belief statement from “strongly disagree” (5) to “strongly agree” (1) to assess the extent to which respondents perceive others in their local community to act in ways that benefit the environment: “Most people in my local community engage in activities that help protect the natural environment” (Larson et al., 2018). Efficacy refers to the extent to which an individual controls their actions and perceives the outcome of their behavior. Environmental efficacy was measured with a single indicator adapted from Oreg and Katz-Gerro (2006), asking respondents to rate the extent to which they agree or disagree with the statement, “There is no point doing what I can for the natural environment unless others do the same.” Both items were recoded as binary variables for inclusion in regression models.

We divided sense of place into two measures reflecting natural and civic place attachment (Scannell and Gifford, 2010). Based on confirmatory factor analysis, we used a unidimensional natural place attachment scale (α = 0.88) to measure the extent to which recreationists were attached to the primary place where they recreate. We used four items to measure civic place attachment (i.e., social capital) based on involvement in local community activities and interactions with neighbors to reflect the extent of social ties, providing an affective dimension of social capital (α = 0.84). Items were presented as belief statements on a 5-point scale from “strongly disagree” (1) to “strongly agree” (5). Responses were averaged to construct composite indices where higher values reflect greater place attachment/social capital and lower values indicate less.

Past engagement in stewardship activities was measured as a set of pro-environmental behaviors adapted from previous research (Cooper et al., 2015; Larson et al., 2018) to reflect the wildlife conservation context of this study and presented as behavioral statements assessing the frequency with which respondents participated in various stewardships activities in the past 5 years on a five point scale: never (1), rarely (2), occasionally (3), often (4), or very often (5). Our unidimensional scale included eight items (α = 0.85) and averages were computed for each respondent to create a continuous composite index where higher totals (on a scale of 1–5) indicate more frequent engagement in stewardship activities and lower totals indicate less frequent involvement.

Additionally, we included a variable to gauge awareness of current conservation funding mechanisms. Several options were presented, along with the most accurate statement, “Funds from the sale of hunting and fishing licenses and equipment,” and a binary variable was created to indicate a correct answer (1), or incorrect answer (0). Membership in a conservation organization was measured as a binary variable, where (1) indicated the respondent belongs to an international, national, or local conservation or environmental organization, land conservancy, hunting-related organization, or birding/wildlife watching organization, while (0) indicates that they do not.

A specialization construct was measured using seven items representing respondent behavior, skill, commitment, and centrality to lifestyle of their primary recreational activity (α = 0.80). Items were combined and summed into a specialization index on a scale from 10 to 44 with higher values reflecting greater levels of specialization.

Socio-demographic information was collected from respondents, including age, gender, ethnicity, income, education, type of residence, and region of the state, and included as control variables in regression models and cluster analysis. Age, income, and education are continuous variables, while gender, type of residence, and region of the state were coded as dichotomous dummy variables. The reference category for each binary demographic variable is indicated in Table 2. Ethnicity was not included as a category for analysis due to the predominately homogenous sample of respondents indicating a white ethnic background (91%), and 6% of respondents choosing the “no response” option.

We conducted all statistical analyses using Stata 14 statistical software (StataCorp, 2015). We assessed descriptive statistics to compare overall support for various conservation policy options and utilized chi-square independence tests to test for significant differences in conservation policy support between hunters and wildlife watchers. Bivariate correlations (Pearson's r) and Variance Inflation Factor (VIF) values were checked to assess multicollinearity among predictor variables in our regression models (Pearson's r < 0.6; mean VIF = 1.41). We modeled support for conservation funding policies using four binary logistic regression models with each policy as a dependent variable and hypothesized covariates included as independent variables. Additionally, we performed a k-means cluster analysis to segment respondents according to their patterns of responses to the conservation funding policies presented in the survey and explore associated typologies that may be meaningful for wildlife managers and policymakers.

Of those in our overall sample (n = 3,500), 73% identified primarily as hunters (n = 2,558) and 27% as wildlife watchers (n = 942). Respondent age was skewed toward older age groups. The greatest proportion of respondents for both groups was in the 50+ year age range and a mean age of 55 years for hunters and 59 years for wildlife watchers. Hunters were predominantly male (92%), while wildlife watchers were represented by more females (60%) than male (37%) respondents. Both groups reported to be mostly from a white ethnic background (91%) and tended to be college-educated (62%), although wildlife watchers had a greater proportion of post-baccalaureate degrees (27%) than hunters (14%). The most common residential context for hunters (41%) and wildlife watchers (42%) was in suburban or small towns, while more hunters (35%) were from rural areas than wildlife watchers (28%).

Support for conservation policies varied by policy type (Figure 1). Reallocation of sales taxes or lottery proceeds to fund conservation strategies received greater support than expanding the existing “user-pay” model. The idea that extractive industries who profit from natural resources should contribute a portion of their annual revenue to conservation received the highest support, with 88% somewhat supporting or strongly supporting the proposition and only 12% responding neutral or opposed. A user-based tax was evaluated to gauge support for a tax on general outdoor gear (i.e., a “backpack tax”) and proceeds earmarked for fish and wildlife programs (similar to the allocation of Pittman-Robertson funds). This idea received evenly split support, with 42% of respondents opposing or strongly opposing the idea and 41% in somewhat or strong support. The number of respondents “strongly opposed” to a backpack tax was much higher than the other three policies at nearly 27%; 17% reported being neutral on the idea of a backpack tax. The two policies assessed that would provide funds reallocated from state sales taxes or lottery proceeds received relatively strong support, with 72% and 77% in support, respectively. 12% somewhat or strongly opposed the sales tax option, while 13% opposed the lottery proceeds option, with 17% and 11% neutral, respectively.

Differences emerged when we compared support among the two stakeholder groups surveyed (Table 1). Hunters supported the extractive industry policy (86%), but not as strongly as wildlife watchers (93%) (X² = 37.4, p < 0.001). Hunters were also more likely to oppose reallocation of sales tax revenue, with 32% opposed compared to 22% of wildlife watchers (X² = 31.4, p < 0.001). Wildlife watchers supported the lottery proceeds option slightly more than hunters (80% vs. 75%) (X² = 37.4, p < 0.001). Hunters were slightly more supportive of a backpack tax (42%) than wildlife watchers (37%) (X² =7.77, p < 0.01). When we compared awareness of conservation funding mechanisms, 87% of hunters correctly identified revenue from the sale of hunting licenses and equipment as the primary source of conservation funding, compared to only 51% of wildlife watchers (X² = 526.2, p < 0.001).

Binary logistic regression results show how the odds of supporting the given policy (backpack tax, lottery proceeds, sales tax, or extractive industry) change based on the associated covariates (Table 2). Coefficients in logistic regression are log odds which are typically converted to odds ratios for easier interpretation. Odds ratios describe the change in the likelihood of the binary outcome associated with a one-unit change in continuous predictor variables and compare the odds of the binary outcome at two different levels for categorical predictor variables (Hosmer et al., 2013). Therefore, we interpret the odds ratios in terms of overall statistical significance and draw comparisons between models based on patterns observed in the independent variables.

Descriptive analysis and chi-square tests suggested that hunters and wildlife watchers differ in their support for conservation funding options. However, when accounting for the effects of other covariates, participation in hunting or wildlife watching did not have a statistically significant influence on stakeholder support. Regression results indicated that no variables were significantly associated with support or opposition across all four conservation funding policies presented, although several variables were associated with three out of four policies. Odds ratios >1 indicated that more frequent involvement in stewardship activities was associated with increased odds of supporting the backpack tax, sales tax, and extractive industry options. Belonging to a conservation organization did not appear to influence support for any of the policies, and knowledge of conservation funding mechanisms only affected (increased) support for the backpack tax.

Mutualistic value orientations positively influenced support for three policies: lottery proceeds, sales tax, and extractive industry revenue. A domination orientation toward wildlife was associated with decreased odds of supporting the sales tax and extractive industry policies. Attachment to one's primary recreational area was only associated with support for the sales tax option. Conversely, social capital was associated with decreased odds of supporting three policy options: backpack tax, lottery proceeds, and sales tax. However, closely related to the idea of social capital are descriptive social norms or the perception that others in one's local community are engaged in actions to help protect the natural environment. Social norms were positively associated with the lottery and sales tax options, indicating increased odds of supporting those policies when one perceives others in the community as taking pro-environmental actions. However, neither environmental efficacy nor specialization were measurably associated with any of the conservation funding policies.

Demographic variables exhibited varying degrees of predictive influences on support for conservation policies. Age was negatively associated with lottery and sales tax options, indicating that older respondents were less supportive of those particular conservation funding mechanisms. Gender was associated with degree of support for lottery and backpack tax policies, yet the relationship differed for each. Relative to the reference class male, those who identified as female, non-binary, or other had decreased odds of supporting a backpack tax and increased odds of supporting the lottery proceeds option. Education positively influenced support for backpack tax and sales tax options. Neither type of residence (rural, suburban, or urban) nor region of the state of Michigan were significantly associated with support for any of the four conservation funding policies. Hosmer-Lemeshow and Pearson chi-square goodness of fit statistics show p-values above the 0.05 cutoff, indicating that the models fit the data well.

Hierarchical clustering using Ward's linkage identified an optimal number of clusters (3) by iteratively minimizing the distance to the mean from the individual data points in that cluster, followed by a k-means cluster analysis specified with three respondent clusters: “strong support,” “mixed/opposed,” and “anti-backpack tax” (Figure 2). The respondents in these groups were more similar to each other in their patterns of responses to policy variables than they were to respondents in other clusters.

Respondents in the “strong support” cluster comprised 36% of the sample and generally somewhat or strongly supported all four policy options. Regarding the backpack tax, which was the least supported option for the entire sample, the “strong support” cluster exclusively somewhat or strongly supported the option, with zero respondents reporting neutral or opposed responses. The “anti-backpack tax” cluster comprised 43% of respondents, all of whom somewhat or strongly opposed the backpack tax option (with 29% neutral) while generally supporting the other three options. No one in the “anti-backpack tax” cluster supported the backpack tax option. Respondents in the “mixed/opposed” cluster comprised 21% of all respondents; a greater proportion of this cluster opposed policies than supported, except for the extractive industry-based option, and had a higher proportion of “neutral” responses to all four policies.

One-way analysis of variance (ANOVA) for continuous variables and chi-square tests for categorical variables indicated several associations between respondent clusters and covariates (Table 3). Respondents in the “strong support” cluster were more likely to have engaged in stewardship behaviors (Mean = 2.98), a higher percentage were aware of conservation funding mechanisms (81%), and 67% reported being a member of a conservation organization. Those in the “strong support” cluster were slightly more mutualistic in their value orientation toward wildlife (Mean = 3.59) than the “mixed/opposed” cluster (Mean = 3.07). They also had the highest proportion of respondents in the <40 age range (15.8%) and were more likely to have a Bachelor's (33%) or graduate/professional (20%) degree.

Respondents in the “anti-backpack tax” cluster were similar on many variables to the “strong support” cluster, with two notable exceptions: “anti-backpack tax” cluster respondents were less likely to report being a member of a conservation organization (55% vs. 67%) or correctly identify current conservation funding mechanisms (74% vs. 81%). The “mixed/opposed” cluster was slightly less mutualistic and more utilitarian in their perspectives and value orientations toward wildlife than the “strong support” or “anti-backpack tax” clusters. They were slightly less engaged in stewardship activities than the other two clusters (Mean = 2.24) and were the least likely group to report membership in a conservation organization at only 46%. The “mixed/opposed” group also had the highest proportion of male (82.3%), middle-aged (40–64; 65.6%), and rural (34.5%) respondents, though type of residence (urban/rural) and region of the state were not statistically significant.

Several limitations affect the generalizability of our findings. First, our sample comprises individuals in the MDNR e-mail database and likely are not representative of the general population of Michigan or broader geographic areas. Inferences should be limited to the population from which the sample was drawn. Furthermore, our sample consists of wildlife-associated recreationists who may be personally affected by some of the proposed policies (e.g., “backpack tax”) and not others (e.g., lottery and extractive industry options.) Future research efforts might strengthen these findings by drawing on broader sample frames to account for differential impacts of policies on different populations. Second, the conservation funding options presented are not exhaustive and reflect choices based on space and time limitations of the survey. For instance, we did not investigate voluntary programs such as specialized license plates, tax checkoffs, or conservation stamp purchases. Future research would benefit from exploring various other options that exist for funding conservation that may vary in efficacy or political feasibility. Third, no information was provided to respondents regarding the potential implementation of the policy options or the expected impacts on conservation funding. Our study design captured a cross-sectional snapshot of stakeholder support for specific policies; given the dynamic nature of politics, further research may benefit from longitudinal designs or choice experiments that measure how respondents weigh trade-offs and alternatives or how support changes over time in response to new information and social norms. Low R² values in our regression models indicate that we did not adequately measure all factors that account for variance in conservation policy support. Future research could strengthen our model of conservation support by incorporating other potential covariates to create a more comprehensive framework for assessing and predicting conservation policy support. Finally, the data for this study were collected during the COVID-19 pandemic of 2020. It is currently not known how the unprecedented situation may have influenced peoples' responses to surveys.