Impact of Local Empowerment on Conservation Practices in a Highly Developed Country

Community‐based conservation, where local decision makers are responsible for balancing conservation and development, is often preferred to exclusionary conservation that prioritizes use‐limitation through strict regulation. Unraveling the evidence for conservation impact of different governance regimes is challenging. Focusing on conservation practices before and after a reform can provide an early indication of behavioral changes acting as a precursor to changes in social and ecological outcomes, which generally need more time to materialize. A recent reform in Norway provides a unique opportunity to evaluate the impact of local empowerment on conservation practices in protected areas. We analyzed 1,466 decisions in 31 protected areas before and after the reform while accounting for differences between private and public property ownership. We found that the conservation practices were liberal both before and after the reform. The impact of local empowerment on conservation practices was contingent on land tenure: more use was allowed after the reform on private land. We conclude that conservation impact evaluations could benefit from a before‐and‐after spatial approach taking into account land tenure for analyzing the impacts of local decision making.


Introduction
The conservation impact of protected areas depends on their governance, but the effect of different governance regimes on conservation remains uncertain (Macura et al. 2015). Exclusionary conservation where power is held with national authorities and enforced through strict legal regulations is, on the one hand, thought to favor conservation through use-limitation (Bruner et al. 2001;Locke & Dearden 2005), but too strict or poorly devised rules that fail to account for social impacts can reduce local support and rule compliance over time (Agrawal & Chhatre 2007;Chan et al. 2012). Community-based conservation attempts to address these challenges by combining local development with conservation, including stakeholders as active participants and/or devolve control over natural resources (Robinson & Redford 2004;Brooks et al. 2013). Local involvement is expected to gain support for conservation, foster stewardship, and provide both ecological and social benefits through better use of local knowledge in policy implementation (Ribot 2002;Van Laerhoven 2010;Brooks et al. 2013).
The term community-based conservation represents various forms of local involvement depending on the sociopolitical context. It includes cases where: (1) the government grants decision making to local governing bodies, (2) the local communities own or have usage rights in the conserved area due to collective land tenures, and (3) the "local residents exercise de facto control in the absence of formal rights" (Poteete & Ostrom 2004;Hausner et al. 2012). Community-based conservation could further be understood as a bottom-up process where decision making starts at the local level and involves interactions at multiple levels (Berkes 2006;Baral 2012). This perspective reflects the fact that local decision making is constrained by conservation policies and rules originally crafted at higher levels of governance and depends on financial support from external sources.

Figure 1
Conservation objectives of Norwegian protected areas. The purpose of Norwegian protected areas is to safeguard areas of vulnerable and threatened nature, cultural heritage and cultural landscapes, and give the public the opportunity to experience nature through simple recreation. Here, illustrated by pictures of wild reindeer, recreation, and livestock grazing. Photos top, middle-right, and bottom by Morten Kielland and middle-left by Jørn Eriksson.
Other institutions, such as land tenure, add to the complexity of protected area governance (Ostrom 2007). Land tenure comes with a different set of values than conservation-values that are tied to the control of, access to, and extraction of resources (Berge 2006), rather than to the preservation of natural values for the general public associated with conservation ( Figure 1). Society has generally been organized around these institutions for a much longer period than protected areas, providing them with a high degree of local legitimacy. Protected areas in Norway are designated on both private and public land, including land managed as commons, and a recent study has found that land tenure is more important compared with protected areas in determining how local residents value the landscape (Hausner et al. 2015). The above suggests that land tenure could be an important factor affecting the impact of protected area governance.
A recent reform toward community-based conservation of Norwegian protected areas provides a unique case for studying the impact of governance using a beforeand-after spatial design. Establishing protected areas in Norway has resulted in local resistance because residents feel their rights to use, access and decide are, or will be, limited by national authorities. To reduce local conflicts and improve integration between conservation and development the government decided, in 2009, to transfer decision-making authority to local protected area boards composed mostly of elected local representatives. The reform distributed the main responsibility for protected areas over a larger number of decision-making bodies operating at a local scale ( Figure 2). The reform is too recent to evaluate the impact of local empowerment on ecological or social outcomes, but it is possible to isolate the effect of governance on behavioral outcomes, such as the decisions made by local protected area boards regarding the use of protected areas. By investigating conservation practice before and after the reform, the effect of local empowerment can be evaluated at an early stage.
Our main research question relates to whether conservation practices of decision makers have changed because of local empowerment. To answer this question we analyzed conservation decisions associated with permits. Permits are the main tool for regulating activity in Norwegian protected areas and the most immediate way that local boards can influence the level of use. Each protected area has a set of rules (protection regulations) tailored to the local conditions which stipulates the activities that are prohibited, restricted, allowed, or require permits (Hausner et al. 2017). By investigating permits, we can assess which activities are deemed most threatening to conservation and how strictly the protected areas are protected. Ultimately, permit decisions reflect interactions between the different levels of governance, because decisions made by local boards that are perceived incompatible with conservation objectives can be appealed or simply overturned by national authorities. Stakeholders can also appeal permit decisions if they disagree, reflecting their degree of acceptance of the decisions.
We ask: (1) Has local empowerment led to a more liberal conservation practice (i.e., are more permits granted)? (2) Has local empowerment led to an increased acceptance of conservation decisions? (3) Is the impact of local empowerment contingent on land tenure or protected area designations?

Before-and-after spatial design and analysis
Before-after studies are vulnerable to influences from confounding factors, which could be causing the observed difference apart from the intervention (i.e., local decision making; Ferraro 2009). Adding control sites, such as protected areas that match the characteristics of our study sites but where centralized governance has continued, is difficult because the reform has been extensive and includes most national parks (34 out of 37). Using a spatial design to control for other governance influences, such as private and public land ownership, could reduce the risks of confounding factors in the before-after analysis. We analyzed 1,466 decisions made before the reform by the government (i.e., the regional environmental authorities) and after the reform by local protected area boards (see supplementary material for more information about the reform, a description of the study areas and details on data collection). The 31 protected areas included in this study ( Figure 2, Table S1) make up ß20% of the protected land area (kilometre 2 ) of the mainland. National parks (IUCN II) make up 75% of the area, protected landscapes (IUCN V) 24.5%, and nature reserves (IUCN Ia) 0.7%. The proportion of public land is 68% in National Parks, 43% in Protected Landscapes, and 58% in Nature Reserves. The decisions analyzed span a 9-year period, from 2006 to 2014 (Table S2).
We used mixed logistic regression to test the conservation impact of local empowerment. We looked at the probability that an individual application was granted (GRANT), with decision-making authority (REFORM) as the predictor variable. Other variables that could determine the probability of a successful application were included as covariates. These were the different local protected areas boards included in the study, shown in Figure 2C and D (AREA; included as a random factor), the activity applied for (CATEGORY; since some activities conflict more with protection objectives than others and are therefore more likely rejected), PRO-TECTED LANDSCAPE specifying whether the application involved an activity in protected landscapes or not (since protected landscapes allow more activities than national parks and nature reserves) and TENURE (as different property ownership can affect the decision maker's room for manoeuver; see Table 1 for details about model terms).
The response variable y ij , for application i in AREA j was denoted 1 for a granted application and 0 for reject. The probability of a granted application (Pr [y ij = 1]) is p ij . Since the effect of the reform is the main focus of the analysis, the full model contained all the two-way interaction terms between REFORM and the covariates. We fitted a random intercept model with a logit link, with AREA as a random intercept, a i .
We performed model selection using single-term deletion minimizing the AIC. We inspected the model fit from simulated residual plots with values simulated both at the population level (i.e.,without the random effect) and also taking into account the random effect, and tested for overdispersion. For the analyses, we used the software R (R Development Core Team 2016) and the libraries lme4, DHARMa, piecewiseSEM, AICcmodavg, and blmeco (Bates et al. 2015;Korner-Nievergelt et al. 2015;Hartig 2016;Lefcheck 2016;Mazerolle 2016).

Model selection
The most parsimonious model was PROTECTED LANDS-CAPE + CATEGORY + TENURE + REFORM + REFORM: TENURE (see Table 2 for model output). The model was not overdispersed (dispersion glmer = 0.66). Marginal pseudo-R 2 (proportion explained by the fixed factors) of the final random effects model was 0.16 and conditional pseudo-R 2 (proportion explained by the both fixed and random factors) was 0.19. The main effect of REFORM and the interaction between REFORM and TENURE were statistically significant (Table 2).

Conservation practices before and after the reform
We found that the local protected area boards allowed slightly more use after the reform, but this effect was restricted to private land ( Figure 3, Table 2). The conservation practices were liberal, in the sense that most permit applications were granted, both before and after the reform. Regional authorities granted 92% of the applications on both public and private land, whereas local protected area boards granted 92% on state land and 97% on private land.
The majority of the applications concerned motorized vehicle use (69%), most of which were on snow-covered ground. Property owners and other rights holders (land owners, cabin owners, usufruct right holders, hunters, fishers, farmers, reindeer herders) were responsible for the majority of the applications (53%) and recreation and tourism (trekking associations, extreme sport actors, sports clubs, dog clubs, tourism facilities) was the second most dominant stakeholder group (23%). Different categories of use were stricter than others. Figure 3 shows that the probability of a rejected application was highest for activities like buildings, industry development, and motorized vehicle use on bare ground.
The probability of allowing an activity was higher (marginally significant) for applications that concerned activities in protected landscapes compared with national parks and nature reserves. In total, 43% of all applications After After reform = local protected area board consisting mainly of local politicians.

Before
Before reform = regional environmental authority is the state's regional branch of the national government that held management responsibility in the study areas before the reform.
AREA Breheimen, Jotunheimen, Nordland, Naerøyfjorden, Reinheimen, Stølsheimen Each local protected area board is responsible for clusters of protected areas, except Stølsheimen that consists of one protected landscape.

Snow, air, and water motor
Motorized vehicle use on snow-covered ground, in the air or on water.

PROTECTED LANDSCAPE
A dichotomous variable reflecting if the application concerns a protected landscape (yes) or a stricter designation (no). The protected areas in the study area consist of national parks (NP), nature reserves (NR), or protected landscapes (PL) as defined by IUCN criteria. An application may concern more than one protected area, and the variable only specifies whether a protected landscape is included or not.
TENURE A dichotomous variable specifying whether the application involves protected areas where the public owns less than (private) or more than 50% of the area (public). In cases where the application concerned more than one protected area, the area owned by the state was summed across these protected areas and divided by their total area, giving an overall proportion of state versus private land for all the protected areas included in the application.
GRANT A dichotomous variable saying if the application was granted or not.
Applications that were granted and partly granted were defined as granted.   Strictness level before and after the community-based conservation reform: The probability of a granted permit-application (±1 SE) before and after (the reform for the different use categories, and by land tenure. Filled and hollow circles separate between before and after, respectively. Plot A and B separate between the probability of a granted decision depending on whether more than 50% (public) or less than 50% (private) of the protected area(s) (kilometre 2 ) affected by the permit application was on public land. Estimates are predictions from a mixed logistic regression model when the model term PROTECTED_LANDSCAPE has been defined as no, meaning that probabilities reflect applications that involve nature reserves and/or national parks (see text for details).
concerned activities in a protected landscape even though protected landscapes only make up 24% of the total area. The number of permits for buildings and industrial development relative to the total area was higher in protected landscapes than in national parks ( Figure S1). National authorities only appealed or overturned local board decisions on five occasions. Stakeholders also seemed relatively pleased with the decisions since they appealed a slightly lower portion of permit decisions after the reform (3.95% before compared with 2.50% after the reform). Permit applications that concerned buildings, motor use on bare ground, and industrial development were appealed most often relative to the number of applications in these categories (Table S3).

Evaluating the impact of local empowerment on conservation practices
The conservation impact of local empowerment is challenging to evaluate empirically. Site comparisons are frequently used to evaluate whether community-based conservation or centralized governance is more effective in terms of attaining conservation impact. The use of before-and-after spatial designs that allow for the assessment of alternative explanations of the pattern observed is less common (Nolte et al. 2013;Macura et al. 2015). A before-and-after design allowed us to control for the influences of protected area designation and land tenure to detect the impact of local empowerment on conservation practices. The analyses of the reform showed that conservation practices were slightly more liberal on private land after the reform. Conservation is a national undertaking and restricting use on public property is often considered more legitimate than on private property. Conflicts with property owners was one of the main reasons for embarking on this reform, and we found indications of a higher acceptance of conservation practices as a result of local empowerment, at least as suggested by the slight reduction in the number of appeals on conservation decisions.
Most studies on community-based conservation are from developing countries (Brooks et al. 2013;Macura et al. 2015). In highly developed countries, such as Norway, fewer people rely on local resource utilization for their livelihood, and conservation conflicts are more often rooted in the right of self-determination over properties owned through generations (Hausner et al. 2015). Our results suggest that local decision making could alleviate conflicts, but at the expense of more use of protected areas designated on private property. Aichi target 11, aiming at conserving 17% of terrestrial areas and inland waters means that more private land is likely to be conserved in the future, resulting in similar conservation challenges as in Norway.
A premise of community-based conservation is that communities should receive socioeconomic benefits from conservation, either directly by linking conservation with development or indirectly through compensation (Brooks et al. 2013). Because environmental authorities before the reform already granted most uses, it is at present, difficult to see how conservation impact can be improved through local empowerment, since there seems to be little room left for local boards to increase benefits to local stakeholders. There are limits to how lenient it is possible to get without turning the protected areas into "paper parks"-i.e., protected areas that are protected on paper but not managed nor enforced in practice and Norwegian protected area governance is already less strict than other countries with similar socioeconomic and ecological conditions (Fauchald et al. 2014;Hausner et al. 2017). Preferably, impact assessments should be performed before embarking on widespread reforms. This way it is possible to better understand where the opportunities to achieve conservation impact are, and institutions can be designed with a clearer assumption on what to expect from governance interventions.
The lack of any substantial change in conservation practices also points to the fact that reforms often tend to build on each other rather than replace each other (Driessen et al. 2012) and that including counterfactual scenarios of continued centralized governance and "no protection" is needed to better establish causal relationships and assess conservation impact. The next steps should be to analyze avoided use by comparing protected sites with sites that match in all other aspects related to use, but which are not protected. One way to do this is to extend our analysis of conservation practices to compare protected areas with sites subject to municipal planning outside protected areas. Another more common approach is remote sensing to assess long-term ecological impact of protected areas (e.g., Nelson & Chomitz 2011). A limitation of before-and-after spatial design using quantitative analysis of conservation decisions is that we cannot rule out a possible change in behavior of the applicants, for example, through better communication with park managers during the decision process. According to a few other studies there are, however, few signs that conservation practices have changed much in terms of involvement of local stakeholders in decision making (Overvåg et al. 2015;Hovik & Hongslo 2016).