Attitudes towards biodiversity conservation and carbon substitution in forestry: a study of stakeholders in Sweden

Global change has fueled debates on forest use and management, including the need to adapt management to mitigate future risks in forestry. Additionally, forests are important for biodiversity cons ...


Introduction
Managed forest is the ultimate example of a socio-ecological system (Blanco et al., 2017).Environmental conditions and societal demands with implications for how the forest is used and managed are key drivers of this system.More recently, the need to use the forest for climate mitigation has emerged through not only carbon sequestration in standing tree volume (storage) but also through carbon substitution, whereby products that either directly or in the production process cause carbon emission to be replaced with wood products (Canadell and Raupach, 2008;Solberg et al., 2017;Trivino et al., 2017).With the opportunities and uncertainties presented by global and climate change, discussions regarding forest management center around increasing carbon sequestration (through carbon storage or carbon substitution) (Bellassen and Luyssaert, 2014), reducing forests' vulnerability to disturbances (Jactel et al., 2017) and biodiversity conservation (Lagergren and Jönsson, 2017).Changes to forest management systems will have social as well as ecological consequences.Increased carbon sequestration through substitution might result in intensified forest management for biomass production that could pose a threat to biodiversity.Whereas carbon sequestration through increasing the standing volume of trees (storage) aids conservation goals, production goals of managed forests can be impeded.Meanwhile, more 'close-to-nature' methods are assumed to increase biodiversity and reduce forest vulnerability to disturbances (Björkman et al., 2015;Felton et al., 2016;Klapwijk et al., 2016;Jactel et al., 2017;Peura et al., 2018).Thus, while evidence suggests that increased biomass production for carbon substitution provides benefits on a global scale, there may be negative impacts on biodiversity levels on a local or landscape level (e.g.leading to habitat changes and losses) (Söderberg and Eckberg, 2013;Gasparatos et al., 2017;Solberg et al., 2017;Trivino et al., 2017).The act of managing forests for different goalsnot only environmental but also production and social/ culturalmakes continuous trade-offs necessary.Moreover, the struggle over how to best use the forest has further intensified with the recent more vigorous attempts to combat climate change (Klapwijk et al., 2018).
The participation of the general public and more involved stakeholders is crucial in sustainable forest management.By increasing the potential for learning and trust, participation can reduce conflicts and facilitate the development of legitimate policy and management (Kozak et al., 2008;Kangas et al., 2010;Eriksson, 2018).In practice, participatory processes are often challenging and not always successful, for example due to contextual barriers or a poorly designed process (Reed et al., 2018).By depicting grounds for disagreement between interest groups and revealing why certain management strategies are resisted, studies of stakeholders' attitudes may aid in the planning of these processes.Even though psychological factors (e.g.forest values and beliefs) have been found to be more important for attitudes than socio-demographic factors, stakeholder affiliation (e.g.forest owners versus the general public) tends to be a significant predictor of attitudes even after a range of psychological factors have been considered (McFarlane and Boxall, 2003;Eriksson, 2018).Drawing on the cognitive hierarchy model (Dietz et al., 1998), the present study examined relationships between environmental problem awareness, forest beliefs and attitudes towards environmental management in forestry among stakeholders in Sweden.Even though the cognitive hierarchy model suggests that a higher awareness of environmental problems should be linked to positive attitudes concerning the management of these problems (Nordlund and Westin, 2011;Eriksson et al., 2013), the basis for attitudes towards management with potentially mixed environmental impacts operating at diverse geographical scales has not been examined (e.g.management for climate change mitigation, which may reduce biodiversity at the local level).

The cognitive hierarchy model
The present study draws on the cognitive hierarchy model to depict how environmental problem awareness is linked to specific forest beliefs and attitudes (Dietz et al., 1998;Stern, 2000;McFarlane and Boxall, 2003).More specifically, attitude theory stipulates that salient beliefs are important for the formation of attitudes (Eagly and Chaiken, 1993).Whereas beliefs are defined as cognitions or thoughts about some entity, an attitude summarizes evaluations of attitude objects and is either positively or negatively valued.Environmental problem awareness reflects beliefs about to what extent the environment is threatened by human activities, and may reflect environmental problems, such as biodiversity loss or climate change (Nordlund and Garvill, 2002;de Groot and Steg, 2008).Forest beliefs are cognitions about forests specifically and concern different dimensions such as economic (e.g.forest revenues), environmental (e.g.characteristics of the ecosystem including vegetation, animals, soil, water) and social/cultural (e.g.aesthetics, restoration, recreation, accessibility and maintenance) (Nordlund and Westin, 2011;Eriksson et al., 2012;Williams, 2014;Kooistra et al., 2018).When considering climate change, beliefs relating to adaptation and change including openness towards for example new forest management methods, may be valuable additions (cf.Lagergren and Jönsson, 2017).This line of research has confirmed that a higher awareness of environmental problems is linked to a readiness to act pro-environmentally and support pro-environmental strategies (Stern, 2000;Nordlund and Garvill, 2002).In turn, forest beliefs have been found to be related to attitudes towards, for example, forest policy and management (Nordlund and Westin, 2011;Eriksson et al., 2013).

Framing of issues
Attitudes are not only formed based on internal cognitions, but may also be influenced by how the issue is framed.For instance, emphasizing a specific attribute will increase the likelihood that an audience will focus on that attribute when evaluating a message (Druckman, 2001).Even though framing effects have been confirmed repeatedly (Benjamin et al., 2017;Mossler et al., 2017), factors such as the source (e.g. government or business), individual differences (e.g.knowledge and attitudes) and the context (e.g.competing or non-competing) influence the magnitude of the effect (Nisbet et al., 2013;Bertolotti and Catellani, 2014;Covey, 2014;Mossler et al., 2017).With regard to climate change, studies have found that local message frames are more effective than global ones in eliciting environmental action (Altinay, 2017).In contrast, Bakaki and Bernauer (2016) found no effect of a local biodiversity frame versus a global climate change frame on support for forest conservation in the general public in Brazil.Nevertheless, they did not assess potential interactions with, for example, beliefs.Whereas climate change frames have been examined in samples from the general public and students, studies of more engaged stakeholders are scarce.

Environmental management attitudes among stakeholders
Studies have provided robust evidence for differences in values, beliefs and attitudes among forest stakeholder groups (Hellström, 2001;Rantala and Primmer, 2003;Eriksson, 2012;Nordén et al., 2017).For example, Berninger et al. (2009) showed that while forest owners and forestry professionals emphasize economic dimensions, both environmentalists and recreational groups value not only the environment (e.g.biodiversity) but also social attributes (e.g.aesthetics).However, forest owners have been found to also appreciate environmental dimensions and to be in favor of at least certain environmental management measures and programs (e.g.biodiversity conservation, carbon sequestration, the carbon offset market and managing for nature protection), even though they may not always be willing to get involved themselves (Eggers et al., 2014;Dickinson et al., 2012;Miller et al., 2012;Thompson and Hansen, 2013;Cook and Ma, 2014;Håbesland et al., 2016;Polomé, 2016;Kelly et al., 2017;Khanal et al., 2017).While economic interests clearly matter for owners' willingness to implement environmental measures in forestry practices, concern for the environment also plays a part, at least for some owners (Cook and Ma, 2014;Mitani and Lindhjem, 2015;Drescher et al., 2017;Kelly et al., 2017).Thus, while it is not surprising that members of an environmental organization generally favor preservation and conservation in forestry (McFarlane and Boxall, 2003;Urquhart et al., 2017), forest owners display diverse beliefs and attitudes.A better understanding of the cognitive basis of attitudes may help clarify why certain environmental management options are favored while others are not.

The present study
Biodiversity conservation in forestry may involve the implementation of specific measures (e.g.leaving tall trees or stumps during final felling) or more comprehensive management for higher levels of biodiversity (e.g.implementing continuous cover forestry) (Lassauce et al., 2011;Peura et al., 2018).Carbon storage and Forestry replacing fossil fuel with wood productsi.e.carbon substitution represent two approaches to mitigating climate change (Canadell and Raupach, 2008).As the issue of climate change mitigation has emerged in addition to biodiversity conservation (Lagergren and Jönsson, 2017), environmental management in forestry has become increasingly complex.To examine how this complexity is delineated in people's cognitions about environmental management, this study examined predictors of environmental management attitudes among forest stakeholders in Sweden.More engaged stakeholders with a forest ownership profile (labeled owner group) and an environmental/recreational profile (labeled nature group) were drawn from members of interest groups.In addition, a student sample, expected to display a lower level of involvement with environmental management in forestry compared with the membership sample, was examined.The cognitive hierarchy of stakeholders was thus examined in a context in which different, potentially opposing, environmental concerns are evident.
Group comparisons were conducted to assess whether the interest groups display differences in environmental problem awareness, forest beliefs and attitudes in line with findings in previous research (Rantala and Primmer, 2003;Berninger et al., 2009;Eriksson, 2012).Hypothesis (H) 1 depicts group differences in awareness of biodiversity loss and climate change as well as forest beliefs.Compared with the owner group, we expected the nature group to display higher awareness of biodiversity loss and climate change, believe ecological and social dimensions to be more important in forestry, and to be more open to trying new methods in forestry.The nature group was furthermore expected to believe forest production to be less important compared with the owner group.In addition, we hypothesized group differences in attitudes (H2).The nature group was expected to have more positive attitudes towards biodiversity conservation and carbon storage in forestry, but less positive views on carbon substitution than the owner group.
As the framing of issues may influence attitudes (Mossler et al., 2017), the study further examined how a global climate change frame versus a local biodiversity frame influences attitudes towards environmental management strategies.We hypothesized an interaction between environmental problem awareness and frame.Thus, we expected respondents with a high awareness of biodiversity loss to display a more positive attitude towards management for biodiversity in the local frame and respondents with a high awareness of climate change to display a more positive attitude towards climate mitigation (either storage or carbon substitution) in the global frame (H3) (cf.Altinay, 2017).
Drawing on the cognitive hierarchy model (cf.Dietz et al., 1998), we outlined a model depicting relationships between beliefs and attitudes.As environmental problem awareness is believed to be based on more general cognitions such as biospheric values and an ecological worldview (Stern, 2000), the model depicts that awareness of biodiversity loss and climate change are positively correlated (H4).In line with previous research (Nordlund and Westin, 2011;Eriksson et al., 2013), awareness of biodiversity loss and climate change were furthermore believed to be positively related to forest beliefs reflecting an emphasis on environmental and social interests as well as openness to try new methods in forestry, but negatively correlated with an emphasis on production interests (H5).Although previous research has not identified the specific set of forest beliefs underlying environmental management attitudes (but see Nordlund and Westin, 2011), we expected attitudes to be positively correlated with forest beliefs reflecting environmental and social interests as well as openness beliefs, but negatively correlated with forest beliefs reflecting production (H6).Since environmental problem awareness may be directly linked to attitudes according to the cognitive hierarchy model (Stern et al., 1999), direct paths between problem awareness and attitudes (positive relationships) were furthermore considered possible (H7).Even though socio-demographic factors have been found to be less important than psychological factors for attitudes (Eriksson, 2018), we analyzed the importance of gender, age, region and interest group for environmental management attitudes after the psychological factors had been controlled for.
The framing effects and the cognitive hierarchy were further examined using a student sample expected to be less engaged with forest and nature issues than the membership sample.Because lower involvement with an issue may be associated with attitudes that are less stable and thus easier to influence (Eagly and Chaiken, 1993), we anticipated stronger framing effects in the student sample compared with the membership sample.The cognitive hierarchy has been supported among activists and the general public (McFarlane andBoxall, 2000, 2003).Thus, we expected the proposed cognitive hierarchy to be substantiated also in the student sample.

Study context
Forests cover 70 per cent of the land area in Sweden.The majority of the forest is privately owned, with almost 330 000 individual private forest owners owning approximately half of the forest (Swedish Forest Agency (SFA), 2014).The forest is economically important.Sweden had the largest volume of round wood removals in the European Union (EU) in 2015, and is, together with Germany, the EU's leading sawn-wood producer (Eurostat, 2017).In 1993, the production and environmental goals (including an emphasis on biodiversity conservation) were deemed equally important in the Swedish forest policy and this was confirmed in the 2008 government bill, but with the addition that climate change mitigation and adaptation should also be considered (Anonymous, 2008).Because of the importance of forest production, coupled with an emphasis on environmental protection in the forest policy, both environmental and non-environmental concerns are salient in the forest debate in Sweden (Zaremba, 2012;Backman and Mårald, 2016).Comparable to other countries, in Sweden forest stakeholders include forest owners, the forest industry, and groups representing various interests such as recreational, environmental and indigenous groups (Rantala and Primmer, 2003;Berninger et al., 2009;Sténs et al., 2016).However, even the general public may be considered a stakeholder because of the importance of the forest in terms of a national asset (Eriksson, 2018).

Participants
For the membership sample, an owner group representing interests in private forestry and a nature group with either environmental or recreational interests were recruited via different non-governmental organizations.A link to the questionnaire was placed on the websites of two of the forest owners' organizations (Södra in the south and Mellanskog in the middle of Sweden), and for the third organization (Norra in the north of Sweden) a group of 263 elected representatives was given the Attitudes towards biodiversity conservation and carbon substitution in forestry opportunity to answer the questionnaire.A group of 380 members in an organization with a focus on recreation (Friluftsfrämjandet) were contacted via e-mail with an appeal to fill in the questionnaire, and a link to the questionnaire was made available on a website accessible only by members of an organization with a focus on nature conservation (Naturskyddsföreningen).
In total, 251 respondents answered the questionnaire.Almost half, 48 per cent (n = 121), were recruited via the forest owner organizations and the rest via the environmental/recreational organizations.The ownership connection was considered important in this study and comparisons were thus made between a group with connections to forest ownership (n = 143) and a group with only connections to environmental and/or recreational organizations (n = 108).Because forest owners in Sweden form a heterogeneous group with confirmed differences in value profiles (Eriksson, 2012), it was unsurprising that the owner group also included members of environmental/recreational organizations (n = 37).The mean age in the owner group was slightly higher than in the nature group (58 years (SD = 12) and 55 years (SD = 15), respectively), and the former contained a smaller share of women than the latter (20 per cent and 47 per cent women, respectively).While the owner group was more evenly distributed between the regions (27 per cent north, 43 per cent middle and 31 per cent south), the nature group included fewer respondents from the north region (14 per cent north, 42 per cent middle and 44 per cent south).
For the student sample, students from different faculties at Umeå University and the Swedish University of Agricultural Sciences (SLU) in Uppsala, Sweden, were invited to participate via an e-mail or an entry on the internal student website (excluding forestry education programs).The sample consisted of 123 students, 80 per cent women, with a mean age of 25 years (SD = 4).While 77 per cent lived in the middle region, 12 per cent and 11 per cent lived in the north and south regions, respectively.The majority were studying in the natural or medical science domain (64 per cent), and those remaining in the social sciences, humanities or behavioral sciences.Although students in forestry programs have not been invited to take part in the study, half of the student sample (n = 62) displayed connections to forest ownership, environment and/or recreation interests (with 28 displaying an ownership connection, 27 a connection to environmental and/or recreational organizations and 7 connections to both interests).

Measures
The questionnaire to the membership sample covered background information (i.e.gender, age and county recoded into regions corresponding to the organizational set-up of the SFA) and measures of environmental problem awareness, forest beliefs and environmental management attitudes (see Supplementary Table S1.for full details).Problem awareness of biodiversity loss [BIO PA] and climate change [CC PA] was measured via three items each.Forest beliefs included a measure assessing openness to new methods in forestry [OPENNESS].In addition, the importance of eight aspects in forestry was assessed and an exploratory factor analysis revealed two components (eigenvalues: 3.280 and 1.744) explaining 63 per cent of the variance (see Supplementary Table S2.for factor loadings).The components were labelled eco-social beliefs [ECO-SOCIAL] and production beliefs [PRODUCTION].
Before evaluating nine environmental management options in terms of good or bad (i.e. an attitude, Eagly and Chaiken, 1993), respondents were randomly assigned to either a global climate frame (n = 129) or a local biodiversity frame (n = 122) (see Supplementary Table S1).An exploratory factor analysis revealed three attitude components (eigenvalues: 3.669, 1.515 and 1.001) explaining 69 per cent of the variance (see Supplementary Table S3.for factor loadings).Longer rotation loaded slightly higher on Component 1 (0.560) than Component 2 (−0.521).However, based on concerns for construct validity and reliability, this item was reversed and included in the second component to distinguish between biodiversity conservation [BIO CONSERVATION] (Component 1) and climate mitigation (carbon substitution versus storage) [CARBON SUBSTITUTION] (Component 2).While the alpha of Component 1 was reduced from 0.83 to 0.80, the reliability of Component 2 improved from an alpha of 0.63-0.69.Thus, the attitude among stakeholders towards carbon storage was closely related to that towards biodiversity conservation, but this was not the case when it came to carbon substitution.More forest for storage was the only item loading on Component 3, and was thus only included in the descriptive analyses.
Besides questions regarding educational domain, the same questions were used in the study of the student sample.The internal reliability for the predictor variables was reasonably good, and only production beliefs had an alpha value below 0.70.However, while BIO CONSERVATION displayed high internal reliability, CARBON SUBSTITUTION did not (α = 0.38), indicating that this attitude measure was not coherent in this sample.When longer rotation was excluded the alpha increased to 0.52, and the remaining three items were used in the analysis in the student sample.As the reliability was far from satisfactory, however, this should be considered when interpreting the results.

Procedure and analyses
Via an online web survey, the study of the interest groups was conducted in 2017 and the study of students in 2017-2018, following ethical guidelines for studies in Sweden involving humans.Participation was voluntary and based on informed consent.No personal data (e.g.addresses) were recorded, and no sensitive information was included in the questionnaire.
SPSS Statistics 24 (IBM corp., 2016), was used for analyzing the data, except the path analyses, for which AMOS 24 (Arbuckle, 2016) was utilized.Univariate ANOVAs (including partial eta 2 ) were used for the group comparisons in the membership sample with group as factor and environmental problem awareness, forest beliefs and attitudes as dependent variables, respectively (H1 and H2).To evaluate framing effects (H3), two univariate ANCOVAs were conducted with group, problem awareness (BIO PA and CC PA, respectively), frame (global climate change and local biodiversity, respectively) and the interaction between problem awareness and frame as independent variables.Attitudes (BIO CONSERVATION and CARBON SUBSTITUTION, respectively) were used as dependent variables.Relationships between environmental problem awareness, forest beliefs and attitudes were assessed by means of bivariate correlations (Pearson) and path analyses (H4-7).Because only a few items assessed each concept and the sample size was limited, path analyses were deemed more appropriate than for example structural equation models with latent variables.The maximum likelihood estimation method was utilized to estimate the parameters in the path model.To assess model fit, we used chi-square and the root mean squared error of approximation (RMSEA) (absolute fit indices) and Bentler's Comparative Fit Index (CFI) (relative fit index).We considered an RMSEA value of 0.05 or lower to be indicative of a good fit (Browne and Cudeck, 1993), and the P value of close fit (PCLOSE) reveals whether the RMSEA value significantly differs from 0.05.A CFI value of 0.95 or higher was deemed a fairly good fit (Hu and Bentler, 1999).Finally, two ANCOVAs were conducted to assess the importance of socio-demographics (gender, age, region and group) on attitudes after controlling for environmental problem awareness and forest beliefs.
Descriptive analyses of the three groups in the student sample (i.e.owner group, nature group and no connection) were conducted using univariate ANOVAs (including partial eta 2 ) with group as factor and environmental problem awareness, forest beliefs and attitudes as dependent variables (including pairwise comparisons with Bonferroni correction).Framing effects and the importance of the different psychological factors Forestry were evaluated using two ANCOVAs with group, frame, problem awareness, forest beliefs and the interaction between problem awareness and frame (BIO PA in the BIO CONSERVATION model and CC PA in the CARBON SUBSTITUTION model) as independent variables, and attitudes as dependent variables.

Descriptive results
In line with H1, the nature group displayed a higher awareness of problems, believed eco-social dimensions to be more important in forestry and were more open to trying new methods in forestry compared with the owner group in the membership sample (Table 1).In contrast, the owner group emphasized production in forestry to a greater extent.The owner group displayed fairly positive views on particularly carbon substitution, but also on specific biodiversity measures (i.e.leaving trees/ stumps and dead wood) and increasing the share of forest to store carbon.The nature group was positive towards all strategies, except using shorter rotation as a way to mitigate climate change.Consistent with H2, the owner group was found to be more positive towards carbon substitution, while the nature group was more positive towards biodiversity conservation (Table 1).
The comparison of the owner group and the nature group in the student sample revealed the same differences regarding environmental problem awareness and forest beliefs (Table 1).The group with none of these connections was positioned inbetween.However, whereas the expected group differences were found in the attitude towards biodiversity conservation, the groups displayed no significant differences towards carbon substitution.Overall, group differences in the student sample were less pronounced compared with the membership sample.

Predictors of environmental management attitudes in the membership sample
In contrast to H3, results revealed that neither frame nor the interaction between frame and environmental problem awareness significantly influenced the attitude towards biodiversity conservation or carbon substitution (Table 2).
To examine relationships between environmental problem awareness, forest beliefs and environmental management attitudes (H4-H7), bivariate correlations and two path models of attitudes towards biodiversity conservation and carbon substitution Five-point scale (1 = totally disagree, 5 = totally agree).
2 Seven-point scale (1 = not at all, 7 = to a great extent).Attitudes towards biodiversity conservation and carbon substitution in forestry were analyzed.The bivariate correlations revealed significant correlations between all variables except between eco-social beliefs and production beliefs.Positive correlations were found between problem awareness (biodiversity loss and climate change), beliefs reflecting an emphasis on openness and eco-social dimensions, and the attitude towards biodiversity conservation, but negative correlations with production beliefs (Table 3).In contrast, the attitude towards carbon substitution was negatively correlated with environmental problem awareness, openness and ecosocial beliefs, while a positive correlation was found with production beliefs.
Except for the paths between awareness of climate change and two belief measures (openness and production), the analyses of the two path models revealed that all proposed relationships were significant, though not always in the expected direction.Furthermore, the model fit was poor for the biodiversity conservation model (df = 5, χ 2 = 128.190***,CFI = 0.823, RMSEA (PCLOSE) = 0.314 (0.000)) and modification indices suggested a direct path between awareness of biodiversity loss and attitude.After this path was added and the two non-significant paths were removed, the model fit improved significantly (df = 6, χ 2 = 10.403,CFI = 0.994, RMSEA (PCLOSE) = 0.054 (0.387)).
The paths between openness beliefs and attitude, and between production beliefs and attitude, were no longer significant, but explained variance in attitude still increased from 47 per cent to 68 per cent.Compared with the biodiversity conservation model, the model fit for the carbon substitution model was slightly better, although the RMSEA value was significantly greater than 0.05 (df = 5, χ 2 = 29.784***,CFI = 0.955, RMSEA (PCLOSE) = 0.141 (0.001)).Modification indices again suggested a direct path between awareness of biodiversity loss and attitude.With this addition, model fit improved significantly (df = 6, χ 2 = 12.879*, CFI = 0.988, RMSEA (PCLOSE) = 0.068 (0.239)), but the direct effect between eco-social beliefs and attitude was no longer significant.The level of explained variance in attitude increased from 38 per cent to 43 per cent.
The final model shows that in line with H4, awareness of biodiversity loss and of climate change were positively correlated (Figure 1).Whereas the hypothesized relations were found between awareness of biodiversity loss and forest beliefs as suggested in H5, awareness of climate change was only positively correlated with eco-social forest beliefs.Furthermore, the positive correlation between eco-social beliefs and attitude towards biodiversity conservation was in line with H6.However,

Predictors of environmental management attitudes in the student sample
Comparable to results from the membership sample, we found no significant framing effects in the student sample (Table 4).As expected, awareness of biodiversity loss was important for both attitude components; and while eco-social beliefs were important for the biodiversity conservation attitude, production beliefs were important for the carbon substitution attitude.Notably, though, openness beliefs was a significant predictor of only the biodiversity conservation attitude.The explained variance of the biodiversity conservation attitude was comparable to the analysis of the membership sample, but the model explained a considerably lower level of variance in the attitude towards carbon substitution (63 per cent and 17 per cent, respectively).

Discussion
Increased awareness of environmental problems, coupled with a stronger willingness to tackle these issues, has placed forest management at the center of the environmental debate.Using the forest for multiple purposes has generated disputes for some time now (Mori et al., 2017), and when the need to use the forest to mitigate climate change is added, additional demands are placed on it (Klapwijk et al., 2018).This study reveals how stakeholders in Sweden evaluate environmental management in forestry, and the cognitive basis for their attitudes.Overall, results supported hypothesized group differences in the cognitive hierarchy (H1 and H2), but not framing effects (H3) in the membership sample.Whereas expected correlations were confirmed between awareness of biodiversity loss and awareness of climate change (H4), relations between environmental problem awareness, forest beliefs and environmental management attitudes varied depending on whether the focus was on biodiversity conservation or carbon mitigation (thus, only partly supporting H5, H6 and H7).With a few exceptions, group differences in the descriptive analyses and the overall pattern in the cognitive hierarchy were confirmed in the student sample, although framing effects were not substantiated.The student sample displayed weaker ties to forestry and nature interests compared with the membership sample.Nevertheless, the involvement was higher than expected (with only half reporting no connections).Whereas results from the student sample can provide insights on the generality of the cognitive hierarchy, the relatively high involvement level may have impaired the possibilities to observe framing effects.
Adding to previous research of environmental management among private forest owners (Thompson and Hansen, 2013; Polomé, 2016; Kelly et al., 2017), the present study provides insights regarding attitudes towards both climate mitigation and biodiversity conservation, and enables comparisons with stakeholders emphasizing different interests.While the owner group was more positive towards carbon substitution, the nature group favored biodiversity conservation and carbon storage.However, the owner group displayed a positive attitude concerning specific biodiversity measures, such as leaving dead wood and debris in the forest.As the requirement to consider biodiversity when managing forest in Sweden was reinforced already in the 1993 forest policy (Bush, 2010), this may reflect an acceptance of a practice that has been employed for some time now.However, the less positive view of, for instance, forest preservation and continuous cover forestry in the owner group, suggests that they mainly favor biodiversity measures as part of continued forest production, but not management schemes in which production aims are secondary.The nature group generally favored carbon substitution but was less positive towards shorter rotation, suggesting that they endorse the use of wood products for carbon substitution but would not like this to occur via shorter rotation schemes in forestry.Overall, the study reveals two attitudinal standpoints when it comes to environmental management in forestry: support for biodiversity conservation aligned with the attitude towards carbon storage versus carbon substitution.The conflict between biodiversity conservation and carbon substitution identified in the ecosystem and on a policy level is thus mirrored in the cognitions of stakeholders (cf.Söderberg and Eckberg, 2013;Gasparatos et al., 2017).The divergence in attitudes towards different climate mitigation options, carbon storage and carbon substitution, was less obvious in the student sample, perhaps reflecting a decoupling of the evaluations of these measures.Since different sample characteristics, such as a lower level of involvement in forest issues, may have influenced perceptions, evaluations of carbon substitution and storage require further attention in future research.
In the present study, framing in terms of climate change (a global issue) or biodiversity (a local issue) did not influence attitudes towards environmental management, and environmental problem awareness did not moderate the influence of frames on attitudes.The frame was similar to how issues are presented in real-life situations, and may have not been sufficiently salient to influence the evaluation of management.Since no manipulation check was included in the questionnaire to confirm that the frame had been considered, this explanation cannot be ruled out.Another potential explanation is that, because respondents were engaged stakeholders, and even the student sample displayed ties to the interest groups, their attitudes may be stable and not easily modified by different frames (Eagly and Chaiken, 1993, see also Benjamin et al., 2017).While previous research suggests that frames influence public attitudes (Altinay, 2017), there are potentially important moderators (e.g.knowledge) (Mossler et al., 2017); and to provide insight into how to communicate with stakeholders, studies need to examine how frames may have an influence on more engaged stakeholders as well.
In general, the cognitive structure of environmental problem awareness and forest beliefs was found to be important for environmental management attitudes.Nevertheless, the study revealed that the grounds for a positive opinion of biodiversity conservation and carbon substitution in forestry differed.The basis for the positive attitude towards biodiversity conservation was found in the awareness of biodiversity loss and eco-social beliefs, and the high level of explained variance in attitude indicates that the model is a fair representation of the cognitive basis of this management attitude.In contrast, a positive attitude towards carbon substitution stemmed from lower awareness of biodiversity loss, lower openness to new methods in forestry, and a greater emphasis on production in forestry.Thus, comparable to previous studies of forest owners' motives to engage in environmental management (e.g. Kelly et al., 2017),

Forestry
the importance of motives unrelated to the environment was evident also in the present study.The lower level of explained variance in the attitude towards carbon substitution among stakeholders, and the fact that interest group was able to explain an additional 9 per cent of the variance in attitude after the inclusion of socio-demographic and psychological variables, suggest that additional variables are likely important for understanding the basis of attitudes towards carbon substitution among engaged stakeholders.While similar predictors were identified in the student sample, openness beliefs were positively linked to the attitude towards biodiversity conservation but unrelated to that towards carbon substitution, indicating a need to further examine the role of openness beliefs in environmental management attitudes.When interpreting results, there are limitations to consider.Even though different interest groups were examined in the present study, the results cannot be considered to cover all different stakeholder groups in Sweden (as, for instance, indigenous groups are not included).The overall pattern of beliefs and attitudes was in line with expectations, although the non-random sample selection and relatively small samples indicate a need to use caution when interpreting attitudes towards particularly the individual management measures examined in the study.Notably, though, previous research on engaged forest stakeholders has mainly employed qualitative analyses of even smaller samples (e.g.Berninger et al., 2009), and this study thus provides an quantitative assessment of beliefs and attitudes in forest stakeholder groups.By departing from the cognitive hierarchy, the study further contributes to the integration of social science theory into applied forest research.While previous research has examined the cognitive basis of environmental management attitudes among owners, overall model fit has not always been assessed (Cook and Ma, 2014) and the explained variance in the dependent variables has been limited (Drescher et al., 2017).It is however worth noting, that because this study examined attitudes, not intentions or behaviors, forest owners' willingness to personally engage in environmental management was not assessed.To verify results and enable generalization, the proposed model should be tested in different samples, including important interest groups and the general public in different countries.This is particularly important since the models were modified slightly.By exploring the cognitive hierarchy in different contexts varying in the emphasis placed on different management objectives, for example, general and context-specific characteristics of the cognitive basis of attitudes can be explored.
A nuanced understanding of stakeholders' perceptions is needed to improve policy and outreach regarding how to mitigate future risks in forestry and optimize the use of climate change mitigation.Comparable to previous research (Nordlund and Westin, 2011), this study suggests that resistance to biodiversity conservation in forestry may lie in the relatively strong emphasis on production among private forest owners, but also their lower concern for biodiversity issues.As biodiversity conservation and carbon substitution are both desirable goals in Sweden (e.g.SFA, 2017), awareness of stakeholders' conflicting perceptions may help in attempts to gain support for these goals in the future.For example, to increase acceptance for biodiversity conservation among forest owners, there may be a need to more strongly convey the positive link between biodiversity and productivity in forestry (Liang et al., 2016).In addition, when working towards a shared understanding of environmental management in forestry it is important to consider similarities between stakeholder groups (e.g. both groups in this study considered environmental and production interests to be important).

Conclusions
This study revealed a divide between stakeholders' attitudes towards biodiversity conservation and carbon substitution in forestry.While the owner group was more positive towards carbon substitution, the nature group favored biodiversity conservation and carbon storage.Nevertheless, simplistic preconceptions of stakeholders' beliefs and attitudes should be avoided since it is possible to display a negative attitude towards shorter rotation but still favor carbon substitution (cf. the nature group) and to oppose continuous cover forestry but simultaneously approve of specific biodiversity management measures (cf. the owner group).Profound knowledge of stakeholder groups can help policy makers and practitioners to identify a common ground between stakeholder groups.In addition, an understanding of the beliefs underlying attitudes is valuable when preparing for participatory processes with stakeholders and designing outreach.Overall, the opinions of important stakeholder groups highlight key societal demands on forests, and knowledge of these groups is necessary for an understanding of managed forests in terms of a complex socio-ecological system.

3
Seven-point scale (1 = little consideration, 7 = much consideration).4Five-point scale (1 = very bad, 5 = very good).†Item included in the carbon substitution attitude.† †Item reversed and included in the carbon substitution attitude in membership sample only.† † †Item included in the biodiversity conservation attitude.

Figure 1
Figure 1 Environmental problem awareness (biodiversity loss problem awareness (BIO PA), climate change problem awareness (CC PA)) and forest beliefs (openness beliefs (OPENNESS), eco-social beliefs (ECO-SOCIAL), and production beliefs (PRODUCTION)) as predictors of attitude towards biodiversity conservation (BIO CONSERVATION) in Figure 1a and attitude towards carbon substitution (CARBON SUBSTITUTION) in Figure 1b (final models, membership sample).Paths significant at P < 0.05 in bold.Explained variance in endogenous variables (R 2 ).

Table 1
Means, standard deviations and the magnitude of group differences (Partial η 2 ) for environmental problem awareness, forest beliefs and environmental management attitudes in the membership sample and the student sample Means having the same superscript letter did not differ at P < 0.05 (ANOVA in the membership sample and ANOVA with Bonferroni correction in the student sample).

Table 2
Results from the ANCOVA of attitudes towards biodiversity conservation and carbon substitution by group, environmental problem awareness, frame, and the interaction between frame and environmental problem awareness in the membership sample Note.EPA = Environmental problem awareness.†Owner group and nature group (two groups).† †Biodiversity loss problem awareness in the biodiversity conservation model and climate change problem awareness in the carbon substitution model.† † †Global climate change and local biodiversity (two frames).Adj R 2 = 0.66 for biodiversity conservation attitude and 0.33 for carbon substitution attitude.

Table 3
Bivariate correlations between environmental problem awareness, forest beliefs and environmental management attitudes in the membership sample

Table 4
Results from the ANCOVA of attitudes towards biodiversity conservation and carbon substitution by group, frame, psychological predictors, and the interaction between frame and environmental problem awareness in the student sample Note.BIO PA = Biodiversity loss problem awareness, CC PA = Climate change problem awareness, ECO-SOCIAL = Eco-social beliefs, PRODUCTION = Production beliefs, EPA = Environmental problem awareness.†Noconnection group, owner group and nature group (three groups).††Global climate change and local biodiversity (two frames).†† †Biodiversity loss problem awareness in the biodiversity conservation model and climate change problem awareness in the carbon substitution model.Adj R 2 = 0.63 for biodiversity conservation attitude and 0.17 for carbon substitution attitude.