Mass support for conserving 30% of the Earth by 2030: Experimental evidence from five continents

Significance

To combat ongoing biodiversity loss, scientists have called for 30% of land, waters, and seas to be conserved by 2030. Nearly all nations have recently ratified the target, but exceptional efforts are needed from political leaders to achieve it within the short time frame. We study a main determinant of political feasibility, public opinion, of the target in eight countries: Argentina, Brazil, India, Indonesia, South Africa, Spain, Sweden, and the United States. We find that the general public in these countries is strongly in favor of the target, and we experimentally test how policy design can bolster or hinder support. These findings have practical implications for how efficient biodiversity protection can be promoted within the remit of public acceptance.

Abstract

Rapid global expansion of protected areas is critical for safeguarding biodiversity but depends on political action for successful implementation. Following widespread ratification of the Kunming-Montreal Global Biodiversity Framework, an unprecedented increase in area-based conservation is required to reach its target of conserving 30% of land, waters, and seas by 2030. These expansions prompt difficult trade-offs between conservation, social, and economic interests. A key factor in securing legitimacy and practical feasibility for expansion regimes is understanding what factors determine public support for them. Using survey and experimental data, we show that in eight countries across five continents, public opinion is 1) strongly in favor of the “30-by-30”-target and 2) highly consistent regarding policy priorities for the design of international- and domestic-level expansion regimes. We find that for international-level policy regimes, support increases with protection responsibilities equally split between countries, rich countries bearing higher costs, more countries actively cooperating, and placement trade not allowed. For domestic-level policy regimes, support generally increases when nature values are prioritized over social or economic values and, in many countries, decreases when costs are borne by a general tax increase, parks are managed by private companies, and when access to parks is restricted. Together, these results demonstrate how protected area expansion policies can be shaped to facilitate reaching 30% protected areas by 2030.

The Earth faces a rapid loss of biodiversity—endangering the existence of inhabiting animals, plants, and humans. About one in eight species of plants and animals are currently at the risk of extinction (1). A potent remedy is conserving nature in protected areas: geographical spaces with restrictions on human activity. Recent calls by conservationists suggest that at least 30% of land, waters, and seas need be conserved before 2030 (2–5). In December 2022, 195 nations and the European Union^* agreed to this “30-by-30”-target by adopting the Global Biodiversity Framework (GBF) at the COP15 meeting in Kunming-Montreal (6). For the target to be achieved, however, an unprecedented expansion of protected areas is needed; current global protection stands at 17.6% terrestrial and 8.6% marine coverage (7). With less than five years left, over 100 nations are less than halfway in at least one protection category. For example, China and the United States would need to approximately double, and India—quadruple, their terrestrial protection coverage to reach the designated level.

However, protected area expansion entails extensive resource claims, and successful implementation requires navigation of conflicts between conservation, social, and economic interests (2, 8). Efficient biodiversity protection will demand area use restrictions, including cessation of resource extraction and human relocation, affecting significant parts of the global population (9–11). In direct financial costs, annual targeted investments of up to $200 billion may be needed (6, 12), comprising a steep increase from an estimated $24 billion spent in 2020 (12). The multifarious costs (13) and difficult trade-offs prompted by protected area expansion suggest that protecting areas with the highest conservation benefits will not always be possible (14). Instead, policymakers need to seek compromises that are politically feasible.

This study covers one central aspect of political feasibility—public opinion. Public opinion (the aggregate of individual views about a societal topic) shapes the scope of actions available to political leaders. Respect and consideration of citizen preferences are also fundamental ideals in democratic societies, and tend to increase leaders’ chances of staying in power in most political systems (15, 16). Lack of support for protected area expansion can lead to conservation of less biologically valuable areas, or to protection not being expanded at all. Conservation regimes with low public support may also be less long-lived (17), as grievances and promises of change are often used by political opposition to challenge such policies. Conversely, if conservation regimes are designed in ways that garner high support, it means that more effective protection measures could be possible (8). Echoing these concerns, section 7§e of the GBF recognizes “mobilization of broad public support at all levels” an explicit priority (6).

A key tool for political leaders in promoting support and avoiding opposition is policy design (17–20). Both between and within countries, the 30-by-30-expansion actualizes issues where policy choices could lead to diametrically opposite levels of support. A fundamental theme connecting several of these concerns relates to fairness and division of burdens: How conservation and financial responsibilities should be allocated (21, 22). Notably, the GBF provides only minimal guidance in these regards. However, for environmental policy support in general, fairness perceptions have been found to be one of the strongest predictors (23). This suggests that understanding what divisions of burdens are perceived as more or less fair could be central to designing effective expansion regimes.

At the international level, countries need to agree on terms of cooperation for achieving a global expansion of protected areas—including where to place them and how to fund them. This is complicated by countries differing significantly in conservation potential, social costs of expanding protection, and financial capabilities to provide effective management. Recent estimates suggest the costs of a 30-by-30-expansion may fall disproportionately on lower-income countries (24), echoing a history of criticism that area-based conservation projects provide global benefits at the expense of local communities, especially indigenous and marginalized groups^† (25–27). At the domestic level, additional public concerns and corresponding policy decisions arise, including what societal groups are directly affected by expansion (28), how protected areas are managed (29), and what access and use restrictions are put on protected areas (30). A key aim of our study is to investigate how the design of such policy solutions influence opinion (see Supplementary Text, pp. 2-3, for further discussion of specific policy dimensions).

We present cross-sectional and experimental survey data on public support for the 30-by-30-target from 8 countries across 5 continents: Argentina, Brazil, India, Indonesia, South Africa, Spain, Sweden, and the United States. The study provides two central contributions: 1) we measure overall support for a national protection of 30% of terrestrial and marine areas by 2030, and 2) we experimentally test how variations in policy design (trade-offs in policy arrangements and between policy priorities) influence preferences between protected area expansion regimes. This differs from previous research on attitudes toward protected areas which primarily comprise case studies studying perceptions of single and existing protected areas, voiced by communities in the closest vicinity (29–31). These studies generally find positive views of protected areas (32–34), though, by design, cannot reflect the scope of an undertaking at the scale of the 30-by-30-expansion. There is also a need to move beyond cross-sectional research methodology to enable causal inferences about how policy design factors influence opinion. Addressing these gaps is critical to inform researchers, policymakers, and interest groups about what opportunities and pitfalls to expect when rapidly expanding protected areas across the world.

Public Support for the 30-by-30-Target in Eight Countries

We fielded the survey in May 2024; just over five and a half years before the vowed realization of the 30-by-30-target. At the time, Brazil (31% terrestrial, 27% marine protected area coverage) and Spain (28%; 13%) were the countries in our sample closest to achieving the target (7). Three countries were halfway or more with protection on either land or in the sea: Sweden (15%; 16%), the United States (13%; 19%), and South Africa (9%; 16%). Another three countries lacked 18 or more percentage points in both area types: Indonesia (12%; 3%), Argentina (9%; 12%), and India (8%; 0.2%). Countries were selected to reflect wide geographical dispersal, including countries from both the global north and global south, and dispersal of current protected area coverage.

Through market research company YouGov, we recruited nationally representative (or online nationally representative) samples of 1500 individuals from each country (pooled final sample = 12 132, Methods). The survey contained two main tasks: 1) questions about respondents’ opinion of the 30-by-30-target and perceptions of its societal impacts and 2) two conjoint experiments (sometimes referred to as “discrete choice experiments”; 35–37) eliciting preferences between hypothetical expansion policy programs (hereafter: “expansion regimes”). These approaches provide complementary ways to understand public opinion. In the first, absolute levels of support for a national implementation of the 30-by-30-target are assessed. In the second, key policy design factors are experimentally manipulated to enable causal inferences about mechanisms influencing support levels for expansion regimes.

In the conjoint experiment section, respondents evaluated hypothetical expansion regimes presented as “arrangements for how an expansion of protected areas could be managed” (Fig. 1). Expansion regimes comprised principal policy dimensions (e.g., funding, placement) with specific policy positions (“policy dimension levels”) randomly drawn from a predetermined set (Figs. 2 and 3 show shortened dimension and level labels; for details, see SI Appendix, Table S2 and Supporting Text). Respondents’ task was to, across several rounds, indicate a preference between two simultaneously presented expansion regimes. Through the random assignment of policy positions to the hypothetical expansion regimes, conjoint analysis allows us to estimate the causal effect that inclusion of a policy position has on preference for expansion regimes (vs. a baseline value), averaged across policy dimension variations (their average marginal component effects, “AMCEs”; 35, 36). A strength of the conjoint approach is that respondents are required to be discerning and make trade-offs between potentially cherished values (for example, between conservation and social interests). Opinions elicited in conjoint experiments have also been validated as accurate predictors of real-world behavior, including voting in referendums (38).

Fig. 1.

Illustration of the conjoint experiment choice task. Clarifications added in bold.

Fig. 2.

AMCE estimates for the international-level conjoint experiment. Estimates show average marginal component effects with 95% CI, pooled (97 056 observations) and separated by country (~12 000 observations). Estimates are based on Preference for Expansion Regime regressed on binary indicator variables for policy design factors (the top level of each dimension is used as reference category), with SEs clustered by respondent.

Fig. 3.

AMCE estimates for the domestic-level conjoint experiment. Estimates show average marginal component effects with 95% CI, pooled (97 056 observations) and separated by country (~12,000 observations). Estimates are based on Preference for Expansion Regime regressed on binary indicator variables for policy design factors (the top level of each dimension is used as reference category), with SEs clustered by respondent.

We included two separate conjoint experiments: one containing policy dimensions at the international level and one at the domestic level. Each experiment included 4 policy dimensions with 3 or 4 levels (Figs. 2 and 3 and SI Appendix, Table S2). The policy dimensions were based on previously established connections with support for environmental policies, being prominent issues in current debates about protected areas, and for applicability in a diverse multicountry context (See SI Appendix, Supplementary Text). Each respondent completed four rounds of choices between expansion regimes in each experiment, resulting in a combined total of 194 112 rankings of expansion regimes (97 056 per experiment). The preregistrations of the data collection procedure, stimulus materials, and our analytic approach can be found on the OSF (https://osf.io/fqw8u/).

Results

Results show strong support for a national protection of 30% of terrestrial and marine areas (Fig. 4). In the pooled sample, 82.4% are in favor of the target, 11% indifferent, and 6.6% against. Support is strong across all sampled countries, but with clear differences in magnitude. Most notably, Argentinian (87.9% supportive) and Brazilian respondents (90% supportive) skew more positively than average, whereas Swedish (66.3%) and US respondents (71.2%) are less strongly in favor (full descriptive results in SI Appendix, Table S3).

Fig. 4.

Overall support for 30% national protected area coverage by the year 2030. Bars represent responses to the question “What is your overall opinion about the target of 30% protected areas by the year 2030 in [your country]?” “Percentage in favor” represents the percentage of respondents answering 5 or above on the scale. Number of observations: Pooled sample = 11 596, Argentina = 1451, Brazil = 1485, India = 1481, Indonesia = 1482, South Africa = 1469, Spain = 1444, Sweden = 1365, and the United States = 1419.

Regression analyses show that support levels are associated with individual-level factors in generally similar ways evidenced for other environmental policy measures (Fig. 5 and SI Appendix, Tables S4 and S12 provide OLS regressions) (23, 39). Being a woman, being concerned about the environment, supporting income redistribution, and having higher trust in government are associated with higher overall support. However, urbanicity, level of education, and income are not significant predictors of support in most countries, and we find age to be negatively associated.

Fig. 5.

Regression-estimated coefficients predicting individual-level overall support for 30% national protected area coverage by the year 2030. Estimates (unstandardized) are derived from OLS regression predicting overall support for the 30-by-30-target (full output in SI Appendix, Table S4). Variables indicating education, income, and urbanicity are based on binary indicator variables (reference categories: “low”; “rural”). Education categories represent < bachelor degree (“Low”), bachelor degree (“Mid”), and > bachelor degree (“High”). Income categories represent <75% of national median (“Low”), between 75% and 200% of the national median (“Middle”), and >200% of national median (“High”). Attitude variables are based on agreement with 5-point Likert scales, except Trust in government, scale: 1 to 10 (variable details in SI Appendix, Table S15). Estimates are statistically significant when the error bars (95% CI) do not cross the zero-line (note: the coefficient for Age is statistically significant). N = 8643.

Conjoint Experiment Results.

Results from the conjoint experiments are visualized in Figs. 2 and 3 (SI Appendix, Tables S13 and S14 provide the regression output). In the figures, a baseline level is displayed at the top of each policy dimension and subsequent dots indicate AMCE estimates compared to respective baseline value.

The international-level conjoint experiment (Fig. 2) finds that respondents favor expansion regimes where rich countries pay a comparatively higher share of the costs associated with global protected area expansion. Compared to a regime where countries pay their own costs, having rich countries pay more leads to a 4.4% average increase in a regime being preferred (pooled sample). This effect is found across both more and less affluent countries. Similarly, regimes where only rich countries pay are preferred over an equal cost division, almost unanimously across countries, leading to an average 2.7% increase in the likelihood that the expansion regime is preferred.

While expansion regimes with asymmetrical cost distributions are favored, respondents in our experiment generally prefer an equal, and non-negotiable, distribution of the responsibility to protect areas. This is shown in the top and bottom variable sections of Fig. 2: Regimes are preferred where all countries have an equal responsibility to protect areas, and where placement trade (an opportunity for countries to pay to place protected areas abroad) is not allowed. Notable exceptions are Indian and American respondents who are not deterred by placement trade opportunities but prefer them similarly to no trade being allowed.

The number of countries actively contributing to global expansion also matters for expansion regime support. Compared to a baseline scenario where about one-third of the world’s widely recognized countries contribute (65 countries), getting two-thirds or more of the global community (130 countries) to actively contribute makes the regime 3.1% more likely to be preferred by respondents (pooled data). Stepping up to 190 contributing countries is, however, associated with only a modest additional increase of 0.4 percentage point.

At the domestic level, preferences between the included policy positions are generally weaker, and heterogeneity between countries is higher (Fig. 3). The strongest effects are found for what values ought to guide placement and expansion of protected areas. Here, we find a preference for nature values: A regime prioritizing nature is 5.1% more likely to be preferred compared to a regime prioritizing economic values, and 2.8% more likely to be preferred compared to a regime prioritizing social values (pooled sample). The effect vs. economic values is significant in each country, and the effect vs. social values is significant in all countries but Sweden and the United States.

Expansion regimes are generally preferred when funding is provided by taxes on activities harming biodiversity, or when funds are shifted from environmental or welfare investment (as compared to a general tax increase). However, country heterogeneity is evident. Indonesian, Swedish, and American samples show a weaker preference (generally not statistically significant), and for South Africa, adding new taxes to activities harming biodiversity is associated with lower preference for an expansion regime.

Protected area access restrictions, and management, matter less for preferences between expansion regimes in our experiment. Respondents from all countries but Sweden and the United States are indifferent to the full range of restriction levels included. For management, Indian, Indonesian, Swedish, and US respondents are deterred by the possibility of protected areas being run by private companies.

Robustness of Conjoint Experiment Results.

The heterogeneity of these results across respondent subgroups is explored in a large set of additional analyses (SI Appendix, Figs. S1–S16). A notable result is that sensitivity to variations in the policy dimensions included differs between respondents who support and oppose the 30-by-30-target overall (SI Appendix, Fig. S12). In the international-level experiment, respondents opposing expansion are only influenced by how costs are distributed between countries, with expansion regimes where rich countries contribute the most being preferred (similarly to other respondents). In the domestic-level experiment, the same group differs by putting a higher value on lenient restrictions (allowing sustainable commercial activity), and being indifferent between environmental, social, and economic values guiding protected area placement. A similar pattern exists for rural respondents who, compared to other dwellers, put a higher value on all less restrictive access levels in our experiment, and appreciate social values equally to nature values in guiding protected area placement.

We also find evidence of a respondent sophistication effect, in that effects are often stronger in the experiment completed second in order (SI Appendix, Fig. S17; experiment order was randomized). Still, all point estimates consistently appear in the same direction in both presentation orders, meaning that respondents do not switch but hold similar opinions at slightly different strengths (an exception is Management, where two near-zero estimates change direction).

Besides these findings, results are highly consistent across heterogeneity in individual differences, including for gender, age, income, education, urbanicity; attitudes toward income redistribution, environmental concern, collectivism, and trust in government agencies; and perceptions of the societal impacts from protecting 30% of national areas.

Discussion

Successful implementation of the 30-by-30-target will be facilitated by—sometimes dependent on—broad public support. Our study provides two main results in this regard. First, mass opinion is strongly supportive of the “30-by-30”-target in the eight countries sampled. This suggests that while there may be many obstacles to the expansion, public opinion is currently not one of them. Because of the clear patterns of strong support in these countries, this inference should be robust even in light of a plausible degree of measurement imprecision, as sometimes can be the case for public opinion polls (40). Second, views on how an expansion should be implemented are conditioned by policy design factors. Our results suggest that at the international level, support increases by countries dividing the responsibility to protect areas equally, by more countries actively contributing to the cause, and by not allowing countries to pay their way out of domestic protection and place protected areas in other countries. In particular, we find a clear consensus across all countries in our sample that most of the costs needed to achieve the 30-by-30-target should be borne by richer countries. At the domestic level, placing protected areas such that these benefit nature—rather than social or economic concerns—induce less opposition. That is, people do not want “paper parks” but consider the environment fundamental even in trade-offs with other key values. Among those opposing the 30-by-30-target, we find that cost distribution, and access restrictions, are areas where policy design could increase support.

The results provide insights into how public opinion is affected by protected area expansion policy design. The implications can connect, for example, to ongoing efforts of financing global protected area expansion by means of an international fund (41): According to the public in all eight countries we sampled, rich countries should stand for most of the funding. National governments will, however, also need to form national budgets to fund protected areas. Here, our results suggest that funding schemes relying on budget redistributions, or specific taxes on activities harming biodiversity, may find higher support than a general tax increase. Connecting to recent discussion about the benefits of privately governed protected areas (42–45), we also find that this management form, compared to state governance, decreases public support for protected areas in most countries sampled.

Our study’s strengths include its targeted focus on attitudes toward large-scale expansion of protected areas, its inclusion of quantitative data from a diverse set of countries, and its combination of cross-sectional and experimental methodologies. The findings offer insights which could prove valuable to policymakers set to navigate the societal challenges of achieving 30% protected areas by 2030. Scientifically, we extend research on policy support in the environmental domain by studying a wide range of policy design factors, whereas previous research (e.g., on climate change) largely has focused on price-based mechanisms (23, 46). Our cross-sectional results also corroborate findings from previous research about what individual-level factors are associated with environmental policy support and suggest that these determinants operate similarly in the context of support for biodiversity protection by means of protected areas (23, 39, 46, 47).

The study is, however, also subject to limitations, and we have several suggestions for a research agenda that builds on this work.

First, previous knowledge about the 30-by-30-target may not have been widespread among respondents. The results may therefore reflect preliminary opinions that could develop with increasing knowledge and potential politicization of the issue. The relatively small effect sizes found in the experiments could be interpreted as evidence of such naivety [though these magnitudes are comparable to similar previous research; (17–20)]. Notably, however, we find similar reactions to policy design variations in countries already close to 30% protected areas (Brazil, Spain), and others. This suggests that substantial protected area coverage does not necessarily change what policy priorities people prefer from political leaders. Thus, even if some respondent attitudes were formed in a naïve state, it is plausible they could mature without radical change. Future research should nevertheless track the development of political opinion as expansion efforts, and therefore knowledge and potential polarization, ramp up.

Second, we acknowledge that the benefit of designing a survey with a multicountry scope has the drawback of communicating a generalized picture of potential costs and trade-offs to respondents. We urge future public opinion research to study expansion attitudes in multiple local and more contextualized settings, where goal conflicts may be more relatable and reactions to policy design choices therefore possibly stronger.

Third, while our focus is on mass support, we recognize that opposition to the 30-by-30-expansion could come from organized groups perceiving protected areas as harmful. For example, from communities with threatened housing or livelihood, or business interests (25). Future research could add valuable insights by targeting these groups directly, and for instance, study whether compensatory policies could mitigate opposition. Similarly, subgroups in the general public could require specific attention in policy tailoring for support to be upheld. Our findings highlight, for instance, that rural respondents favor a protected area expansion with less strict access levels.

In a broader perspective, it can be noted that the 30-by-30-target has received criticism within conservation science for substituting quality of conservation with mere quantity of conservation (48–50). Naturally, biodiversity protection will require multiple policy measures, and future research ought to extend the present work by taking a public opinion approach to studying these. One imaginable path is to explore “second best” policy options (51), which may have larger political feasibility if they impose less severe use restrictions and therefore create lower levels of local opposition (52). This could include the use of Other Effective area-based Conservation Measures (OECMs), that are geographically demarcated yet not designated a formal protected area status (53). Additional promising avenues for future research include investigation of country-level determinants of progress toward the 30-by-30-target, and of further individual-level factors related to protected areas and biodiversity protection (for example, regarding the extent to which supportive attitudes translate into behavioral outcomes; 54, 55).

In conclusion, designing viable expansion policies is a necessary and urgent step for political leaders striving to reach the 30-by-30-target. Failure to realize a large-scale protected area expansion may come at dire costs, in the form of habitat destruction and continued biodiversity loss. Our results show that people across eight diverse countries are supportive of the 30-by-30-target and that policy design choices can facilitate or hinder favorable public opinion toward protected area expansion. The high consistency of these findings suggests that public opinion may be favorable, and may be similarly influenced, across a wide variety of additional countries. Therefore, as judged by the general public shortly before the final stretch to 2030, this would suggest that promotion of protected area expansion is a politically feasible pursuit.

Materials and Methods

Ethical Considerations.

The study received ethical approval from the Swedish Ethical Review Authority (2023–06835–01). All respondents provided informed consent to participate in the study.

Respondents and Sampling.

Respondents were recruited through online panels maintained by YouGov (www.yougov.com) or local affiliates. Data collection commenced in the second half of May 2024, approximately simultaneously for all sampled countries. Samples were recruited to be representative, or online representative, with regard to age, sex, region, and education level (only age, gender for Indonesia; age, gender, and region for South Africa). Applying demographic weights to analyses does not change interpretation of the results unless otherwise stated, SI Appendix, Fig. S20 visualizes a main regression model comparison. Demographics of the final sample are displayed in SI Appendix, Table S1.

Procedure and Measures.

Respondents answered an online questionnaire consisting of, in order: 1) general individual difference measures, 2) first conjoint experiment, 3) second conjoint experiment, and 4) questions on perceptions of and support for the 30-by-30-target. The primary measure of the final section, overall support for the 30-by-30-target, was measured last to increase the saliency of potential trade-offs and costs associated with implementing the target (that is, to decrease naivete in opinion responses). Presentation order of the conjoint experiments was counterbalanced, such that approximately half of respondents completed each one first.

Before starting the conjoint experiments, respondents were generally informed about protected areas and presented the current percentage of terrestrial and marine protected areas in their country. Specifically, we stated that “Next we will ask you for your opinions about national parks, nature reserves, and other forms of protected areas. Protected areas are geographical spaces created for the purpose of protecting biological diversity (‘biodiversity’). They provide beneficial environments for animals, plants, and ecosystems, by putting limitations on human activity. In protected areas, businesses can typically operate only under strict sustainability regulations, or they are not allowed to operate at all. For individuals, hunting of animals and picking of plants is generally strictly limited or completely forbidden. Sometimes, access to parts of a protected area may not be allowed at all.” Immediately after reading this information, respondents were required to confirm their understanding by answering a comprehension check (see below).

In keeping with the focus on political feasibility, the individual difference measures we included focused on political perceptions and preferences (and environmental concern). After completing the conjoint experiments, respondents answered four questions on perceptions of societal impacts of implementing the 30-by-30-target, as well as the question about their overall opinion of the target. All items are listed in SI Appendix, Table S15.

In addition to the quality assurance measures taken by YouGov, we included two comprehension checks. The first verified respondents’ understanding of “what is meant by a ‘protected area’ in this survey,” and the second pertained to the instructions of the conjoint experiment task. The checks were posed in multiple choice formats, and only respondents passing both checks are included in the presented results.

Conjoint Experiments.

For each conjoint experiment, respondents chose between 4 pairs of conjoint profiles (no ratings were collected). That is, respondents were shown a total of 8 profiles each from the total of 108 (international-level experiment) and 192 (domestic-level experiment) unique profiles made possible through the random combination of attribute levels. No restrictions on attribute-level randomizations were applied. All respondents completed both experiments, and presentation order was counterbalanced between individuals. The experiments were introduced individually, and respondents were instructed to treat them as separate. Details on conjoint attributes and levels are found in the Supplementary text and SI Appendix, Table S2. Respondents were displayed a similar table as part of the instructions but with the “Policy dimension levels” column omitted. Attribute order was fixed for all respondents, as listed in SI Appendix, Table S2.

Data Analysis.

As preregistered, we took an exploratory approach to analyzing the conjoint experiment data. Our analyses focus on estimation of average marginal component effects (35, 36), estimated using the cjoint package for R (35). In these analyses, the dependent variable (Preference for expansion regime, coded 0 for a regime not chosen, 1 for chosen) was regressed on binary indicator variables for all policy design factors, with the first value (SI Appendix, Table S2) used as reference category. All estimates are therefore interpreted as relative to their corresponding reference category. SEs were clustered at the individual level, and α = 0.05 (two-tailed) was used to infer statistical significance.

In analyzing Overall Support for the 30-by-30-target, we used standard OLS regression and ran analyses in R (56). Variable treatments are presented in a note to SI Appendix, Table S4. Applying survey weights makes no meaningful change to the outcome (SI Appendix, Fig. S20).

Visualizations were created using R packages ggplot2 (57), cjoint (35), cowplot (58), and jtools (59).

Deviations from Preregistrations.

Data from a ninth country, Kenya, were collected. Due to low panel availability, the same level of data quality could not be guaranteed by YouGov. We received a total of 1003 completed surveys before comprehension check exclusions were applied. After exclusions, only 269 remained. Because of this, and additional data quality concerns, we have moved reporting of these results to the SI Appendix (see SI Appendix, Figs. S18 and S19 and Tables S16–S19).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data, and code to reproduce analyses, have been deposited at the OSF, and can be found at https://doi.org/10.17605/OSF.IO/FQW8U (60).