What are Canadians willing to pay for bird habitat conservation? A contingent valuation approach

The full questionnaire included questions concerning respondents’ attitudes, behaviours, and preferences for habitat and bird conservation, as well as a contingent valuation section. Respondents were asked about their outdoor recreation activities and their conservation donation history. Socio-demographic information collected included age, income, and gender identity (see Supplementary material S1).

After being presented with an overview of the approach to conservation undertaken by NAWMP partners (Fig. 2), respondents were asked about their willingness to make a one-time donation (WTP) of a given value to NAWMP. Voluntary payments (i.e., charitable donations) were selected over compulsory payments as the latter (e.g., Duck Stamps) have historically contributed to the NAWMP, prompting interest in the viability of donations as a source of wetland conservation funding (see Champ and Bishop 2001 for a discussion of the advantages of voluntary donations over compulsory payments). A one-time donation was selected because they tend to yield more conservative and plausible values for the present value of WTP than repeat payments (Brouwer et al. 2008; Myers et al. 2017). To minimize hypothetical bias, we used a dichotomous-choice question (Hanemann 1984; Whitehead and Blomquist 2006) and provided a “cheap talk” statement reminding respondents of their household obligations, and other opportunities to donate to conservation initiatives (Cummings and Taylor 1999; List 2001; Ladenburg and Olsen 2014; Penn and Hu 2019). We also asked respondents to state their confidence in their donation decision using a six-point interval scale (1 = not confident; 6 = confident) after the donation question (as per Champ and Bishop 2001) so that models could be validated by only including responses from those who were confident in their donation decision (Lyssenko and Martínez-Espiñeira 2012).

Donation values were initially determined through discussions during three focus groups (n = 16) with participants from across Canada in April and May 2022. These values were adjusted following a soft launch first wave (n = 94) and again following a second wave (n = 419) of the survey. The final possible donation amounts included $1, $10, $25, $50, $75, $100, $150, and $200. In the first adjustment, the lowest and highest values were changed from a $10 to a $2 minimum and from $150 to a $200 maximum. The second wave showed that the $2 minimum value had not increased the uptake of donations much more than the $10 donation original minimum level (<50%), and so the minimum donation amount offered was reduced to $1. Donation amounts were presented to each respondent randomly; donation amounts were quota-controlled to ensure that each value appeared evenly through the sample.

Respondents who declined to donate were asked why not to identify protest votes (Carson 2000). Respondents who were willing to donate were asked how they would prefer the money to be collected and how they would prefer for it to be spent using lists determined from the focus groups. We included an “Other” category for each follow up question with a free text field. Free text responses were reviewed and recoded into the pre-determined response categories where possible. The order of the response categories for these questions was randomized for each respondent.

To derive estimates of WTP for wetland conservation under NAWMP and to understand factors associated with WTP, we modelled the CVM responses using the non-parametric Turnbull approach and parametric analysis via logit regression (Haab and McConnell 2002; see Supplementary material S2 for details of the empirical approach).

The survey data were extracted and prepared for analysis in R Studio (R Version 4.2.1, R Core Team 2022) using the package “tidyverse” (Wickham et al. 2019). The Turnbull calculation was estimated using Microsoft® Excel®. Logit models and mean WTP were estimated using Nlogit 6.0 (Econometric Software Inc., Greene 2016). Logit models were estimated without covariates (Model A) and with covariates (Models B and C). Socio-economic variables (age, gender identity, income, and location) were added using stepwise inclusion. Given the biases inherent to CVM, our model selection process was guided by identifying the most conservative WTP estimate while controlling for relevant socio-economic variables and exploring associations between WTP and recreation participation. As the number of responses to each socio-economic question varied (each question in the survey was optional), it was our preference to present parsimonious models to maximize the sample size included in the analysis. We adopted a threshold of p < 0.1 for excluding variables to retain marginally significant terms for discussion. Participation in outdoor recreation activities was modelled in two ways: Model B included a binary variable (Y/N indicating previous participation) for each activity (hunting, birdwatching, and gathering) and Model C included an eight-level nominal dummy-coded variable called “Recreation Activities” (levels: never participated, hunted only, gathered only, birdwatched only, hunted and birdwatched, hunted & gathered, gathered & birdwatched, hunted & gathered & birdwatched), with “never participated” as the reference level. For the logit models only, income was converted to a continuous variable from the categories provided in the survey by taking the midpoint of each category, and using $151 000 as the highest amount to represent >$150 000 and modelled both as a linear and quadratic term. Models excluding respondents who were not confident in their donation response (where confidence was <4 on a scale of (1) not at all confident and (6) very confident) and excluding protest votes were estimated and compared to the final models. All monetary amounts are presented in Canadian dollars unless otherwise stated.

Finally, a series of separate post-hoc χ² tests of association were carried out in R (Package: sjPlot, Lüdecke 2022) to examine (i) reasons for not donating, (ii) preferences for how to donate, and (iii) what the money should be spent on by income (four-level ordinal variable as per original data collection: $0–$49k, $50k–99k, $100k–149k, $150k+), age (four-level ordinal variable: 16–25 years, 26–40 years, 41–65 years, 66 years+), gender identity (two-level nominal variable: female, male; note: small sample sizes of non-binary genders meant that these were excluded from post-hoc analyses), JV (five-level nominal variable: EHJV, PHJV—Prairie Parkland, PHJV—Western Boreal, CIJV, PBHJV), and historical hunting, birdwatching, and gathering participation (total number of activities participated in and binary Y/N for each activity). To mitigate for Type I errors from running multiple tests of the same data, Bonferroni-adjusted alpha values were applied to assess significance (alpha value = 0.006; Field et al. 2012). Yates corrections were applied to calculate p values for matrices with <20% of cells with values <5 (McHugh 2013). Effect sizes were assessed using Cohen's w (small effect = 0.1, medium effect = 0.3, large effect = 0.5; Cohen 1988; Volker 2006).

Overall, 32.1% of respondents (n = 746) reported that they were willing to donate to NAWMP (Table 2), which is smaller than the percentage of respondents who reported making a donation to conservation causes 1 year prior to completing the survey. Two-thirds of these respondents accepted the $1 donation value, and almost half of respondents accepted the $2, $10, and $20 donation values. Acceptance rates fell to around one-third for the $25 and $50 donation values, and one-quarter of respondents accepted the $75 and $100 donation values. Between 16% and 20% of respondents accepted donations of $125 or more. To maintain the monotonicity assumption of the non-parametric Turnbull estimate of mean WTP, the $10 (n = 291), $20 (n = 6), $100 (n = 292), and $150 (n = 292) responses were pooled into the lower categories ($2, $75, and $125 respectively). Pooling the responses into the lower response categories reduces the WTP boundary estimates, i.e., generates more conservative estimates. The lower bound non-parametric Turnbull estimate of mean WTP (see supplementary material eq. 1) is $44.46. Assuming a $201.00 maximum donation value produces an upper bound estimate of $54.81. All three logit regression models (see supplementary material eq. 5; Table 3) had negative and highly significant donation coefficients, suggesting that respondents are less willing to donate larger amounts. Parametric estimates of mean WTP for a one-time donation to NAWMP were between $33.59 and $37.98. Model C generated the most conservative mean WTP estimate of $33.59 (±$6.74; Table 3). The confidence intervals for the parametric estimates fall within the bounds of the non-parametric Turnbull estimate of WTP.

Models with covariates (Models B and C) had lower WTP estimates and better explanatory power (R² values), albeit slightly higher AIC and lower log likelihood values, than the base model (Model A). Model fit statistics were similar between Models B and C. Models B and C both revealed significant effects of income and past participation in waterfowl hunting, birdwatching, and/or gathering on WTP. Age, gender, and location variables were excluded from Models B and C for parsimony as the significance levels were far outside of the bounds of our threshold of p < 0.1. Both models found that respondents with higher incomes were willing to donate more to NAWMP. Income was tested using log income and income as a quadratic term, but these variations did not improve on the original models, and so were not included in the final model specifications. When outdoor recreation participation was modelled as binary variables (Model B), birdwatching and gathering participation had statistically significant (p < 0.05) positive associations with WTP, whereas the hunting participation coefficient was marginal (p < 0.1). The birdwatching coefficient was more than three times greater than gathering. Closer inspection of the combinations of participation in different outdoor recreation activities (Model C) confirmed that birdwatching (on its own or in combination with any other activities) was always associated with a higher WTP. Plant gathering had a positive association with WTP on its own, or when combined with birdwatching, but not when combined with hunting. Waterfowl hunting history was not associated with WTP except when combined with birdwatching. This result may be due to the relatively low number of waterfowl hunters in the sample who only hunted, or only hunted and gathered, and should be treated with caution. Respondents who had previously participated in all three activities had the greatest willingness to donate.

A total of 1578 respondents declined to donate to NAWMP. There was significant variation in the reasons for declining to donate (χ² = 3330.50, df = 5, p = <0.001; excluding “Other”). Not being able to afford to pay was the main reason cited for not donating (Fig. 3). The protest vote, ‘‘I do not believe the money will be used effectively’’, was the second most common reason selected by respondents. Of those who originally responded “Other” (118 individuals), 24 respondents were recoded to ‘‘I cannot afford to pay’’ and two respondents to “It is not my responsibility to support the conservation of birds and their habitats in Canada”. The majority remaining in “Other” (92 individuals) stated they supported other non-conservation causes, or had other financial priorities (55 respondents). Eighteen respondents wanted more information or time to research NAWMP before making a decision and several respondents preferred to make non-financial contributions, or might consider donating at another time of year.

Analyses of associations with reasons for declining to donate revealed small to medium effects of income, gender identity, number of activities participated in, and hunting and birdwatching history (Table 4). Lower income earners (up to $49 000) were more likely to choose ‘‘I cannot afford to pay’’; higher income earners ($100 000–149 000) were less likely to choose this. Women were more likely to choose ‘‘I cannot afford to pay’’ and men were more likely to choose the protest vote option. People who had previously hunted waterfowl were less likely to choose ‘‘I cannot afford to pay’’, and more likely to choose either the protest vote option, or that they did not believe birds or habitat required protection, or that it was not their responsibility to support conservation. This last result should be treated with caution due to the small number of responses to this answer combined with the relatively low proportion of waterfowl hunters in the sample. There was no significant variation in reasons for declining to donate by JV, or for those who had previously gathered.

There was significant variation in responses for preferred donation mechanisms (χ² = 62.477, df = 4, p = <0.001; excluding “Other”). There was not a single dominant preference for donation mechanism, although payment via a non-governmental organization and voluntary donations when filing taxes were the most popular options (Fig. 4). Two of 15 responses in the “Other” category were recategorized into “Payment via a non-governmental organization” and “Purchase of a wildlife collectible”. Of the remainder, three people suggested some sort of point-of-sale mechanism, one preferred to donate directly to NAWMP, and the remainder specified responses relating to money type (e.g., cash, cheque, online). There were no statistically significant differences by income, gender identity, JV, or number of recreation activities (Table 5).

There was significant variation in preferences for how donations should be spent (χ² = 144.380, df = 5, p = <0.001; excludes category “Other”). The most popular funding activity was the purchase of new conservation lands and the least was providing incentives to private landowners to carry out conservation on their land (Fig. 5). Only one response in “Other” (n = 14) was recategorized (into “I don't know”). The remaining “Other” responses comprised five people who preferred that NAWMP decide how to distribute the funds, four who would like the money to go wherever it was most needed, several people who wanted the donation to go to all suggested activities, and a handful of ungroupable responses.

Analyses of relationships between socio-demographic variables and willingness to donate revealed small to medium associations with number of activities participated in, birdwatching, and gender identity. Respondents who had never participated in any of the outdoor recreation activities were more likely to not know what they would prefer their donation to go towards, whereas birdwatchers were significantly less likely to choose this option. Men were more likely to support funding science, whereas women were more likely to support funding public education/outreach. Income, age, JV, and whether respondents had ever waterfowl hunted or gathered were not significantly associated with preferences for what donations would contribute towards (Table 6).

We found significant relationships between socio-demographic variables, including age, income, and gender and WTP, reasons for not donating, payment vehicle, and/or what donations should be spent on. In keeping with our expectations, we found a positive association between income and WTP (e.g., Jacobsen and Hanley 2009). Although age and gender identity were not significant in our WTP models, these characteristics were associated with preferences for payment vehicle and what donations should contribute towards. We did not find any relationship between location and WTP or in post-hoc tests, suggesting a broad geographic base of support for NAWMP. This lack of regional variation indicates that national communication campaigns to solicit donations from Canadians may be relevant if tailored to the outdoor recreation activities that people participate in, rather than where people reside.

Our finding that people who engaged in activities that rely on wildlife habitats are more willing to donate to conservation is consistent with evidence in the literature (e.g., Whitehead 1990; Haefele et al. 2019; Casola et al. 2023) and shows the importance of outdoor recreation participation as a foundation for conservation fundraising. Our study indicates that both birdwatching and plant gathering are important factors in willingness to donate to NAWMP. Encouraging engagement in these activities would be an obvious step to broaden support for conservation. There is a rich evolving literature on birdwatcher motivations (Scott et al. 1999; Hvenegaard 2002; Randler 2022), connections to conservation (Cooper et al. 2015), and WTP for birdwatching experiences (Steven et al. 2017). However, there is limited information about gathering experiences in Canada and more research would be useful to better understand the associated connections to and participation in conservation action.

Unlike other WTP/bird habitat conservation studies, we found limited evidence to suggest that waterfowl hunters were more likely to donate than the general public. The association between hunting and WTP was marginal (Model B), but this relationship disappeared when participation in other recreation activities was accounted for (Model C), indicating that hunters were only more likely to donate if they were also birdwatchers. Our results are similar to previous studies that have investigated more general participation in conservation activities, where recreationists who only hunt were found to engage less often in conservation activities than hunters who also view birds or other wildlife (Sinkular et al. 2022). This result may be due to the relatively small percentage of waterfowl hunters in our sample who did not also engage in birdwatching (2%), making detecting an effect of hunting difficult (i.e., Type II error), or it may be that, as our follow-up question for those who declined to donate cautiously indicates, waterfowl hunters are less likely to believe that wetlands and waterfowl require additional conservation support. Hunters are perhaps likely to receive communications about successful duck production to encourage them to take hunting trips, or, given that dabbling and diving duck numbers have been increasing in recent decades (NABCI 2022), this perception may result from hunting experiences. Our results did not suggest that waterfowl hunters were less likely to donate because they already donate to similar causes; it may be that waterfowl hunters consider the Canadian Wildlife Habitat Conservation Stamp as a mandatory licence purchase rather than a donation towards conservation. As waterfowl hunters are known to have heterogeneous practices and preferences (Schroeder et al. 2006, 2013; Moghimehfar et al. 2017; Sainsbury et al. 2024), additional research into the subject of waterfowl hunters’ engagement in and WTP for bird conservation seems warranted.

Importantly, our results suggest an association between wider participation in outdoor recreation activities such as waterfowl hunting, bird watching, and plant gathering on willingness to donate. Birdwatching appears to be particularly key to donation propensity, whether independently or in combination with other outdoor recreation activities. This additive effect reflects previous findings in the literature that suggest a positive association between WTP for conservation and experiences and familiarity with nature (e.g., Martín-López et al. 2007; Ren et al. 2022). While we cannot infer causation, i.e., that wider recreation participation is driving donation WTP propensity as opposed to a pre-existing affinity with nature causing both increased outdoor recreation participation and higher propensity to donate, it is possible that a diversity of outdoor experiences helps to broaden the appeal of nature and that supporting a range of engagements in natural environments (appreciative and consumptive) may help to increase people's connections with and understanding of values associated with those habitats. Our results suggest that people who engage in all three activities have the highest WTP, providing important implications for initiatives aimed at getting people actively engaged with conservation: strategies that promote bird and plant appreciation and foster a general engagement with nature (e.g., eBird, iNaturalist) may lead to contributions directed toward bird habitat conservation. Yet, funders should be mindful of donation fatigue or the possibility of crowding out contributions from existing donors when targeting recreationists that engage in multiple activites.

We identified heterogeneity in preferences for donation payment mechanisms. Canadians chose donating via an NGO roughly as often as donating via taxation, and via wildlife collectibles such as a Duck Stamp approximately the same number of times as donating via a lottery. Therefore, to take full advantage of potential donations from the general public, managers should explore multi-pronged approaches that provide a variety of opportunities and payment mechanisms. While our estimate of mean WTP is around $30, two-thirds of respondents presented with $1 donation values were willing to donate and more than half were willing to donate $2–$24, indicating promise for mechanisms that capture smaller donations, such as on-site donation options at wetland complexes, or point of sale collections in places that sell bird-related products or outdoor recreation equipment.

The Canadian Wildlife Habitat Conservation Stamp, the only mechanism currently available for Canadians to donate directly towards NAWMP goals, garnered support from a large minority (one-fifth of respondents). Shipley et al.’s (2019) analysis of the Audubon Society bird count participants identified that birdwatchers, including non-hunting birdwatchers, do buy the U.S. Duck Stamp, providing empirical evidence that non-hunters choose to support conservation via wildlife collectibles. The Canadian Wildlife Habitat Conservation Stamp currently costs $8.50, well within the average WTP suggested by our results, suggesting an opportunity to take advantage of Canadian willingness to donate. Making charitable donations directly to specific causes when filing taxes is not currently an option available to Canadians, although payroll deductions for charitable donations are commonly used. In the U.S., there are mechanisms in some states for taxpayers to have an amount deducted from their tax refund, pay additional amount to taxes owed, or simply to make a payment of new funds, when they file their taxes. For example, California uses a “Voluntary Contribution Fund” while New York and Virginia use “Voluntary Contributions” and each state has options for taxpayers to direct their donation to wildlife and habitat conservation. The popularity of the “donation via taxation” option—almost as high as donating via an NGO—suggests that this might be a viable longer term option to explore.

Finally, in terms of preference for donation spending, our results suggest strong preferences for conservation on public land and the purchase of new conservation lands, both of which Canadian Habitat JV partners already engage in (e.g., NAWMP 2023). This alignment is a validation of the social license (Hampton and Teh-White 2019) for NAWMP partner actions. The lack of support for providing incentives for conservation activities on private lands suggests that private lands are seen differently than public lands. A lack of support for funding conservation on private land was similarly reflected in our focus groups, which identified issues of trust and transparency around conservation on private lands (H. Harshaw, K. Sainsbury, and P. Schuhmann unpublished data). The idea that ecosystem services of private lands (especially wetlands) provide public benefits has been documented in the academic literature (e.g., Villamagna et al. 2015), but these benefits may be poorly understood by the public. This result highlights a need to communicate the importance of working with private landowners to the public, to avoid a mismatch in expectations.