Understanding consumer behaviour and adaptation planning responses to climate-driven environmental change in Canada's parks and protected areas: a climate futurescapes approach

Parks and protected areas are a global ecological, social and health resource visited by over 8 billion people annually. Their use can yield substantial benefits, but only if a balance between ecological integrity and sustainable visitation is struck. This research explores the potential influence of climate-driven environmental change on visitation to North America's most popular glacier, the Athabasca Glacier in Jasper National Park, Canada. Photorealistic environmental visualizations were used to gauge visitors’ perceptions of environmental change and potential impacts on consumer behaviour. Results suggest that impacts could substantially diminish the site's pull as a tourism destination. Rather than improving visitation prospects, expert-proposed adaptations underestimated the importance of perceived naturalness and contributed to further potential decline. Findings are relevant to protected areas planning and management. They suggest that a natural path to climate change adaptation is the best way to support both ecological integrity and the long-term tourism pull of protected areas.


Introduction
Parks and protected areas (PAs) are an important driver of tourism (Eagles 2014). Global visitation to PAs is estimated at over 8 billion visits a year, and in North America, the economic impact of visitation is estimated between $350 and $550 billion annually (USD) (Balmford et al. 2015). Varied accounting methods make it difficult to quantify the economic impact of visits to Canadian PAs. However, in 2009, Canadian parks generated $4.6 billion in gross domestic product, $2.9 billion in labour income, and $0.3 billion in tax revenues (CDN) (Canadian Parks Council 2011). These estimates highlight why visitation to parks and PAs is important to revenue generation and conservation reinvestment, and is increasingly seen as a benefit to local and regional economies (Eagles 2014).
Despite economic and other benefits (e.g., health and well-being À see Lemieux et al. 2016), visitation can present a challenge when PA use and the profit motive of operators conflict with a mandate to protect ecological integrity (Whitelaw, King, and Tolkach 2014). This relationship is made even more complex by the prospect of future climate change. Lemieux, Beechey, and Gray (2011a) note that as fixed assets established to conserve representative samples of terrestrial and aquatic environments, PAs are vulnerable to climate-induced shifts in biophysical conditions and habitats. Importantly, these shifts will alter both environmental systems and the user experiences upon which visitation is predicated (Archie 2014;Brownlee et al. 2013;G€ ossling et al. 2012).
Despite these sensitivities, Scott, Simpson, and Sim (2012) argue that tourist perceptions and consumer behaviour responses to environmental change are not well understood in the context of climate change. Key knowledge gaps include how visitors might respond to four major types of climate-related impacts, including direct impacts (e.g., changes in seasonality), indirect impacts (e.g., changes to the biophysical environment), mitigation responses (e.g., carbon offsetting), and altered global socioeconomic conditions that may affect tourist mobility (e.g., regional conflicts) (G€ ossling et al. 2012).
In this paper, we draw inspiration from Dann's (1977) pushÀpull travel motivation framework to explore how climate change impacts and adaptations might influence the pull of climate-sensitive PA features. Focusing on the Athabasca Glacier site in Jasper National Park (JNP), we use photorealistic environmental visualizations depicting plausible climate change scenarios to evaluate visitors' perceptions of environmental change. We reflect on what these perceptions reveal about the impacts of climate change on consumer behaviour and the ability of PA agencies to effectively adapt to climate change. We achieve our research objective by examining four related research questions: (1) What perceptions of future environmental change do visitors to the Athabasca Glacier hold? (2) How might future environmental changes at the Athabasca Glacier influence visitors' perceptions and potential consumer behaviour responses at the site? (3) How can a deeper understanding of visitors' perceptions of future environmental change inform climate change adaptation planning in Jasper National Park and protected areas more broadly? (4) How can environmental visualizations be used as a communication, education and outreach tool in efforts to effectively engage visitors and other stakeholders in climate change adaptation planning?
In support of our research objective, we review two bodies of literature. First, using Dann's (1977) pushÀpull visitor motivation framework as a lens, we examine research on visitor motivations. Second, we review research on the use of environmental visualizations as a climate change communication and engagement tool.

Literature review 2.1. Visitor motivations
Drawing on early visitor motivation research (Dann 1977;Iso-Ahola 1982), tourism is understood here as a socio-psychological experience, and visitor motivations as a psychological disequilibrium that prompts and directs the decision to travel (Snepenger et al. 2006). Viewed through this lens, visitor motivations encapsulate a transaction between an individual (or group), including their wants, needs, and desires, and a destination, inclusive of its unique attributes (Kim, Lee, and Klenosky 2003). One early but still popular approach to conceptualizing and measuring this transaction is Dann's pushÀpull framework (Dann, 1977;Kim, Lee, and Klenosky 2003;Klenosky 2002).
Within this framework, push factors describe forces that lead to the initial decision to make a trip. These factors are often operationalized as the socio-psychological drivers that spur the decision to travel (e.g., the desire for escape). By contrast, pull factors contribute to the subsequent decision to select a specific destination, and are therefore closely associated with destination attributes (Klenosky 2002). As the literature on visitor motivations is now quite expansive, we examine studies that are specific to nature-based tourism and PAs, or that are related to climate change and tourism.
In an early study examining Australian visitors to the US, Uysal, McDonald, and Martin (1994) documented push factors among international travellers. Using factor analysis, they showed that travellers exhibited five motivational clusters, including novelty, enhancement of kinship relationships, escape, prestige, and relaxation/hobbies. Despite the unique context offered by PAs, only the novelty factor was significantly higher among visitors who spent time in a PA. In a more recent study, Kruger and Saayman (2010) examined push factors related to visitation at two South African national parks. Through a comparative study design they demonstrated that not all push factors generalize across different PAs. They found that the drive for escape and relaxation, knowledge-seeking, and nostalgia were common to all visitors, while only visitors to Kruger National Park were seeking out a novel experience.
Research has found that certain push factors (e.g., relaxation or escape) are common across different PA contexts (Saayman and Saayman 2009;Uysal, McDonald, and Martin 1994). However, Kruger and Saayman (2010) remind us that not all PA visitation can be traced back to the same initial drivers. This is also true of pull factors related to PAs. When comparing visits to PA and non-PA destinations, for instance, the natural identity and pristine conditions in PAs stand out as a unique motivating factor (Uysal, McDonald, and Martin 1994). Environmental features, such as hiking trails or attributes that are specific to different destinations, have also been shown to shape visitor motivations, as has the presence of rare flora and fauna (Kim, Lee, and Klenosky 2003;Kruger and Saayman 2010;Saayman and Saayman 2009).
One important influence on visitor motivations to consider in a PA context is how climate-driven environmental change may alter pushÀpull dynamics. The link between climate change and tourism is now well established in the literature (G€ ossling et al. 2012;Scott 2011;Scott, Simpson, and Sim 2012), and in some regions, the climate pull-factor "constitutes the resource on which the tourism sector is predicated" (Scott and McBoyle 2001, 70). In Canada, Scott and McBoyle (2001) used a modified version of the Tourism Climate Index (TCI) and showed that the competitive position of Canadian tourism regions could improve as a result of predicted warming, effectively eliminating Canada's current annual tourism deficit. Recent studies focused on US national parks also suggest that peak visitor attendance in some parks is occurring earlier in the year due to increasing spring temperatures (Buckley and Foushee 2012). In fact, warming-mediated increases in visitation are projected for most months in most parks. It is estimated that total annual visits across the US park system could increase between 8% and 23%, and some parks could see a 13À31-day longer visitation window due to warming trends (Fisichelli et al. 2015).
Warming trends offer opportunities related to increased visitation. However, many studies have also exposed the potential (and realized) vulnerabilities of the tourism sector. Scott, Dawson, and Jones (2008) found that climate-mediated snow loss threatened the very existence of a healthy snowmobiling industry in the US Northeast, but presented a lower risk to the ski industry because of a higher adaptive capacity. Archie (2014) examined barriers to adaptation in US mountain communities with a strong reliance on tourism that is facilitated by federal public lands. Results suggested that, although adaptation planning was taking place within the region, a lack of political will and leadership presented considerable barriers to effective action. Finally, examinations of the indirect impacts of climate change, such as impacts on environmental conditions at sites, have revealed important links between ecological integrity, aesthetic quality, and consumer behaviour among PA visitors. Yuan et al. (2006) found that »20% of tourists surveyed in Lijiang, China may not have visited if the glacier was not present. Coral reef health, including the impacts of bleaching, is also regarded as important to the experience and satisfaction of dive tourists at the Great Barrier Reef Marine Park (Zeppel 2012).
It is critical for PA agencies and related stakeholders (e.g., gateway communities, private operators, etc.) to develop proactive adaptation strategies that can minimize detrimental climate change impacts and capitalize on key opportunities (e.g., shifts in seasonal demand). Given that tourism represents a transaction between visitor and destination (Dann 1977), effective planning will require input as to how visitor perceptions may shape long-term consumer behaviour. Despite this need, G€ ossling et al.
(2012) note that knowledge of how climate change might influence tourism is based on "assumptions about the understanding and perception of climate related changes, as well as resulting changes in demand and motivation" (53). In the following section, we suggest that the use of climate futurescapes can help address this gap.

Climate futurescapes
Three-and four-dimensional computer-generated representations of potential landscapes (i.e., environmental visualizations) have been used to develop and communicate land management scenarios across a range of contexts, including forestry, agriculture, and alternative energy planning (Appleton and Lovett 2005; Lange and Hehl-Lange 2005; Lewis and Sheppard 2006). Lovett (2005) used the term futurescape research to describe this field, noting that futurescape studies share a common focus on landscape scale issues, collaborative decision-making, and the use of three-dimensional (3D) digital representations to present potential landscape futures. Findings from futurescape studies have consistently shown that environmental visualizations can enhance participatory processes by making spatial data more accessible, by improving the comprehensibility of future land-use plans, and by making participatory processes more interactive (Lewis, Casello, and Groulx 2012;Schroth, Pond, and Sheppard 2015).
Over the past decade, environmental visualizations have also garnered growing attention as a climate change planning tool (Bishop et al. 2013;Schroth, Pond, and Sheppard 2015;Sheppard et al. 2011). Research suggests that effective climate change communication and engagement requires messaging that is local, emotional, social, and supportive, indicating that less may be more when it comes to catalyzing public action (O'Brien and Wolf 2010;Groulx et al. 2014;van der Linden 2015). In this regard, Sheppard et al. (2008) argue that environmental visualizations can make "the global both local and personal, putting scientific information into understandable forms and contexts" (4). The high apparent realism contained in environmental visualizations can also evoke strong emotional responses from viewers, and can be leveraged to reflect the social implications of climate change impacts (Lovett 2005;Schroth, Pond, and Sheppard 2015). Finally, as powerful visualization platforms continue to merge with environmental simulations, Lewis, Casello, and Groulx (2012) suggest it may become possible to create data-rich virtual environments that enable deeper expert and non-expert collaboration, ensuring that adaptation planning supports local values, as well as reducing risk (O'Brien and Wolf 2010).
While their potential is widely recognized, the capacity of climate futurescapes to contribute to collaboration and sound decision making depends on their diligent development and use. Visualizing climate change at a landscape scale can be difficult due to the uncertain nature of predicting impacts (Shaw et al. 2009). Climate futurescapes should be based on credible scientific evidence of potential environmental change (Sheppard et al. 2011). However, the uncertainty associated with climate change means that the visualization process inevitably requires many decisions that are far from probabilistic in nature (Downes and Lange, 2015;Lewis, Casello, and Groulx 2012). An effective and justifiable use of environmental visualization technology blends what Tierney (2014) refers to as a possibilistic perspective on risk, with a transparent, credible and systematic (i.e., defensible) visualization process (Sheppard et al. 2008). The blending of possibilistic thinking and a defensible process can ensure climate futurescapes enhance knowledge, collaboration, and concern (Bishop et al. 2013;Schroth, Pond, and Sheppard 2015), rather than contributing to mistrust, misunderstanding, and continued inaction.
Studies have started to borrow concepts and processes from the field of scenario planning to develop a defensible process for creating and disseminating climate futurescapes. Rather than relying solely on quantitative probabilities, qualitative downscaling techniques are used to link possibilistic visualization scenarios to sound scientific evidence (Dockerty et al. 2006;Shaw et al. 2009). In a study examining future agricultural landscapes in the United Kingdom, Dockerty et al. (2006) produced visualizations of potential changes to agricultural land uses. To ensure these visualizations were credible, they integrated a local Climate Land Use Allocation Model (CLAUM) with a policy downscaling exercise that translated national climate scenarios into locally relevant climate narratives. More recently, the Local Climate Change Visioning Project (LCCVP) developed a downscaling framework for creating and visualizing climate scenarios (Sheppard et al. 2011). The starting point for the LCCVP downscaling process is a set of four future climate worlds that present different emissions and response pathways (Deep Sustainability; Efficient Development; Adapt to Risk and Responsibility; Do Nothing À see Shaw et al. 2009). The LCCVP process has been used to visualize and communicate several community-level climate impacts, including flooding and snow line retreat (Cohen et al. 2012;Sheppard et al. 2011). In all cases noted above, visualized scenarios were used to communicate potential impacts to stakeholders, and results showed that climate futurescapes can enhance knowledge, generate concern, and improve levels of engagement around local climate change issues.

Study site context
Canada's 46 national parks were established to protect and present outstanding representative examples of natural landscapes and phenomena that occur across the country's 39 natural regions (Parks Canada 1997). Together, Banff, Jasper, Yoho and Kootenay national parks exemplify this goal as part of the Canadian Rocky Mountain Parks World Heritage Site, which is designated by UNESCO for its exceptional natural beauty and cultural significance. In combination with three other parks in the Canadian Rocky Mountains, these parks capture 59% of the 13.5 million total visits to national parks in Canada (Parks Canada 2015a). In addition to fostering visitation, the Canada National Parks Act (Government of Canada 2000, Section 8(2)) is also clear that " [m] aintenance or restoration of ecological integrity, through the protection of natural resources and natural processes, shall be the first priority of the Minister when considering all aspects of the management of parks." Along with this broader context, several site-and park-specific attributes made the Athabasca Glacier site a suitable case study for this research. Jasper National Park is the largest (11,228 km 2 ) of all the Rocky Mountain Parks and contains a community of 4,700 people (the town of Jasper) that is a cornerstone of the regional tourism industry (Parks Canada 2015c). The site itself has also undergone observed environmental change as a result of climate change. The Athabasca Glacier has retreated approximately 1.5 km, has lost approximately 50% of its thickness (Hugenholtz et al. 2008;Parks Canada 2015b), and is anticipated to almost completely disappear by 2100 due to continued warming trends (in the range of 1.8 C to 2.7 C by mid-century compared to a 1971À2000 baseline) (Clarke et al. 2015;Murdock et al. 2013). The surrounding landscape has also undergone change, including the growth of vegetation in the glacial foreground and the opening of a hollow that is now developing into a proglacial lake (Gadd 2011). Finally, JNP and the Athabasca site are both highly visited destinations. The 2014 JNP annual report estimated more than 2.02 million visitors in 2013, and projected 2.15 million  Figure 1 shows the Athabasca Glacier site in relation to JNP, as well as the location of key visitor experience opportunities.

Development of climate futurescapes
We followed best practice to develop climate futurescapes that were used to assess potential impacts of climate change on visitor perceptions and consumer behaviour. More specifically, we adapted the LCCVP narrative downscaling approach to visualize '2050 scenarios' depicting potential impacts and potential impacts with adaptations. The starting point for our downscaling and visualization process was the LCCVP 'Do Nothing' and 'Adapt to Risk' climate worlds. The 'Do Nothing' climate world that corresponds to our impacts scenario assumes a high-carbon future with little effective climate change action. The 'Adapt to Risk' climate world that corresponds to our impacts with adaptations scenario is based on the same emissions profile, but assumes adaptation efforts are enacted to respond to impacts (Shaw et al. 2009). Introducing a climate world with a different emissions profile (e.g., efficient development) could have covered a broader range of uncertainty in our final set of climate futurescapes. However, Clarke et al.'s (2015) projections for deglaciation in the region found that "until mid-century (»2050), the fate of all glaciers in this area is virtually independent of the emission scenario and climate model used for the projections" (2).
As the LCCVP process was not contextualized to our study, we modified our scenarios during the first phase of our visualization process. Our downscaled local narratives incorporated relevant global and regional climate projections Murdock et al. 2013), and drew on glacial change studies to derive a suite of potential sitespecific impacts. The list of potential impacts included changes in glacial mass-volume, snow cover, debris cover, foreground vegetation, and the potential development of a proglacial lake and stream system (Luckman and Kavanagh 2000;Hart 2006;Hugenholtz et al. 2008;Tennant and Menounos 2013). Work from Clarke et al. (2015) was particularly important in developing a defensible estimate for glacial retreat. Informed by the literature on climate change and tourism, a number of adaptation options were also developed to respond to these potential impacts. These included the introduction of a footbridge and roped fence, the extension of current walking paths, and the adaptation of roads that currently support snocoach tours. While these adaptations sought to respond to multiple needs, overall they prioritized continued safe visitor access to the glacier.
The second phase of our visualization process involved an expert review process. Draft narratives describing potential impacts and adaptations were translated into an initial suite of environmental visualizations. These representations were created in Adobe Photoshop CS5 using photographs of representative viewpoints and a photomontage technique (Sheppard 2001). Ultimately, six images (three scenarios shown from two viewpoints) were prepared. They depicted current conditions (i.e., photographs), impacts, and impacts with adaptations (see Figure 2). The narratives and environmental visualizations were reviewed by three glaciologists and twelve tourism and climate change experts. This expert review proved critical to refining several assumptions in our initial narratives and environmental visualizations. For example, comments from glaciologists suggested that it would be very difficult to depict changes in debris cover in a defensible manner. As a result, debris cover was calibrated to current conditions. Comments from climate change and tourism experts also indicated that our proposed adaptations did not anticipate a wide enough range of potential adaptation options. Given their use at other glacier sites, the addition of helicopter tours was recommended.

Data collection
Surveying for our study took place at the Athabasca Glacier site during a high-volume visitation period between the 24th and 31st of August 2014. Surveying efforts were distributed over weekdays and weekend days and surveyors were onsite consecutively from the morning until the mid-afternoon (roughly 9:30 am until 3:30 pm most days). Data were collected using tablet computers and the iSurvey/DroidSurvey software. Recent research has shown that tablet computers yield equivalent results to traditional pen and paper surveys, and have the added benefit of reducing participant fatigue (Ravert, Gomez-Scott, and Donnellan 2015). The iSurvey/DroidSurvey platform allowed us to include our visual stimuli in the survey, although the screen size of our tablets (Google Nexus: 10.1 00 and Apple iPad: 9.7 00 ) limited the level of detail depicted. To overcome this potential limitation, we printed larger (11 00 £ 17 00 ) image booklets and made them available to participants (see Online Supplementary Materials).
Potential participants were approached at the Parks Canada information kiosk (located near the Athabasca Glacier hiking trail) after they viewed the glacier. Participation was solicited on a next available basis, meaning that an interested visitor and one of two surveyors were available to start the survey. Surveying across weekdays and weekends helped to promote greater participant diversity. When a potential participant over the age of 18 showed interest in completing a survey, a description of the study purpose, survey procedure, and processes for ensuring voluntary participation and data confidentiality was provided. This description was offered verbally and was outlined in more detail in an information letter that was provided. When potential participants consented to participation, they completed the self-administered survey on a tablet computer. A member of the research team was available at all times to address any questions. A chance to win a $100 gift card to an outdoor retailer was offered as a survey incentive.

Survey design
Our survey included three separate sections. The first section covered questions about the nature of each participant's trip and their general and place-specific perceptions of climate change. Climate change perception questions for this section were included in response to the call from G€ ossling et al. (2012) to examine consumer responses to climate impacts within a tourism context. General questions about climate change were developed from past studies examining public perceptions of climate change (Roser-Renouf et al. 2014;van der Linden 2015), including questions about the existence of climate change (1 D very sure climate change is not happening; 5 D very sure climate change is happening), the cause of climate change (1 D mostly human activities; 2 D mostly natural variations in the environment), and participants' level of concern for the climate change issue (1 D not at all concerned; 5 D extremely concerned). As visitation to the Athabasca Glacier involves direct visitor experiences with climate change impacts, we also asked questions that gauged participants' perceptions of climate change as it relates specifically to the site. We assessed whether they believed the Athabasca Glacier would disappear due to climate change (1 D disagree strongly; 5 D agree strongly), how long they believed this might take (open response), and whether they would consider visiting another location to view glaciers (1 D no; 4 D definitely).
As we were interested in consumer responses to possible future environmental conditions at the site, a considerable portion of our survey involved participants evaluating and rating visual scenarios depicting plausible future landscapes. This second section of the survey included three scenarios that are shown in Figure 2. For the current conditions scenario, participants rated their landscape preference (1 D dislike a great deal; 5 D like a great deal), perceived naturalness (1 D not at all natural; 5 D completely natural), and likelihood of a return visit (1 D extremely unlikely; 5 D extremely likely). Using the same response scales, participants also rated their landscape preference, perceived naturalness, and likelihood of a return visit for the 2050 impacts scenario and 2050 impacts with adaptations scenario. For these two future scenarios, visitors rated a second consumer behaviour question that asked whether they would have made their current trip if the landscape conditions they expected to experience were similar to those being depicted (1 D no; 4 D definitely). Based on best practice in environmental visualization research, we collected ratings for multiple (i.e., 2) viewpoints and averaged these ratings for analysis (Lange 2001;Lewis 2012). In combination with our multiple landscape scenarios and multiple questions per image, this requirement yielded 22 stimuli ratings overall.
The third and final section of the survey included key visitor demographic questions covering age, citizenship, gender and education (see Online Supplementary Materials for questions and responses used in the digital survey).

Data analysis
Our analysis approach was twofold. Descriptive statistics were produced in SPSS (v.21) to characterize general perceptions of climate change and specific perceptions of local climate change impacts. Inferential and correlational analyses were organized around our environmental visualization scenario rating exercise. These analyses added depth to our descriptive results and linked participants' perceptions of environmental change to potential impacts on consumer behaviour (i.e., visitation). We followed Tabachnick and Fidell (2013) and assessed normality using a range of indicators. As deviations from normality were not severe, we used parametric tests.
Repeated measures ANOVA tests with planned contrasts were used to compare ratings of the different visualized scenarios. Effect sizes were calculated using the Cohen's d statistic for all planned contrasts. For the impacts with adaptations scenario specifically, a correlational analysis also examined how visitor-level characteristics (e.g., education, perceived naturalness, etc.) related to the desire to visit the Athabasca Glacier site.

Visitor and trip characteristics
A total of 252 surveys were collected from adult visitors to the Athabasca Glacier. The response rate of 49.5% was acceptable, although variable weather conditions were a factor. Twelve participants were removed during data cleaning, leaving a final analysable sample of 240 visitors. With the exception of four items that contained between 5% and 14% missing data, most survey items contained fewer than 5% missing values and many contained less than 1% missing values. Little's MCAR test indicated that values were not missing systematically. Due to the result of the MCAR test, the small proportion of missing data across most variables, and our sample size, missing data were addressed through listwise deletion (Tabachnick and Fidell 2013).
Our final sample was composed of a near equal proportion of male (51.2%) and female (48.8%) visitors ranging in age from 18 to 74 (mean D 36.7) years. Nearly half (49.1%) of the sample had at least a Bachelor's degree. A total of 63.7% of the sample was composed of non-Canadian citizens and over half (58.3%) of the sample resided outside of Canada. Trips among Canadian residents predominantly included visits from Alberta (n D 40), Ontario (n D 19), and British Columbia (n D 18). Most participants (76.7%) were first-time visitors, and on average those who had visited before had not done so in many years (mean D 17.4; sd D 11.23). Trip lengths to JNP were distributed across groups who did not stay overnight (16.7%), who stayed one night (22.6%), and who stayed multiple nights (60.7%; mean D 3.7). Overall, sampled visitors indicated that they were satisfied with their experience at the glacier (mean D 3.71, sd D 0.85).

Perceptions of climate change and environmental change
The vast majority of participants (92.0%) took the position that climate change is occurring. Within this overall majority (i.e., among the 92.0%), most participants (79.3%) also took the position that climate change is caused primarily by human activities. A similar sized majority (70.9%) was either moderately or extremely concerned about the climate change issue. Responses to more specific questions about glacial change reflected a similar pattern. A large majority (82.5%) of participants either agreed or strongly agreed with the statement: "The glacier will disappear from Jasper National Park due to climate change." On average, visitors indicated that this might happen sometime shortly before 2100 (median D 80 years).
When asked whether they would be willing to view glaciers somewhere other than JNP, most visitors suggested they would probably or definitely be willing to do so (83.3%). None of the above results were significantly different between first-time and return visitors. Results were also not significantly different between Canadian residents and non-residents, with the exception of the final statement. Compared to residents of Canada (mean D 2.91; sd D 0.72), non-residents (mean D 3.17; sd D 0.66) were significantly more willing, on average, to view glaciers elsewhere (t D 2.81, p < 0.01, d D 0.38) (see Online Supplementary Materials for additional detail on this section).

Environmental visualization scenario results
Participants rated their landscape preference, perceived naturalness, and likelihood of return visitation using the same item and response options for all three scenarios shown in Figure 2. As such, we assessed whether ratings were applied reliably across the three scenarios. Cronbach's alpha levels indicated that visitors used the landscape preference (a D 0.81), perceived naturalness (a D 0.84), and likelihood of return visitation (a D 0.85) scales reliably across the three scenarios. Figure 3 shows these ratings (averaged across the two viewpoints). Landscape preference, perceived naturalness and likelihood of return visitation were highest for the current conditions scenario, lower for the impacts scenario, and lowest for the scenario presenting impacts with adaptations.
Results of a repeated-measures ANOVA test, shown in Table 1, indicate that rating differences across the three scenarios were significantly different for all three measured variables. Planned contrasts further indicate that ratings for landscape preference, perceived naturalness and likelihood of return visitation were also significantly different between all pairs of scenario comparisons. These differences yielded medium to large effect sizes for ratings of landscape preference, smallÀmedium to large effect sizes for perceived naturalness, and smallÀmedium to large effect sizes for likelihood of return visitation.
For each landscape scenario, we examined relationships between visitors' perceived likelihood of a return visit, their landscape preference, and their perceived naturalness. The perceived likelihood of a return visit was moderately or strongly correlated with landscape preference for the current conditions (r D 0.472, p < 0.001), impacts (r D 0.622, p < 0.001), and impacts with adaptations scenarios (r D 0.687, p < 0.001). The perceived likelihood of a return visit was moderately correlated with perceived naturalness for the impacts (r D 0.482, p < 0.001) and impacts with adaptations scenarios (r D 0.458, p < 0.001), and somewhat weakly correlated with perceived naturalness for the current conditions scenario (r D 0.304, p < 0.001).
Given our focus on planned adaptation, a more detailed examination of the likelihood of visitation under the impacts with adaptations scenario was of particular interest. Table 2 depicts an analysis of factors that correlated with the likelihood of return visitation, and the likelihood visitors would have made their current trip if they expected their experience to reflect the conditions shown. As noted above, the likelihood of a return visit under this scenario was correlated with visitors' ratings of landscape preference and perceived naturalness. Landscape preference and perceived naturalness were also strongly and moderately correlated with the likelihood that visitors would have made their current trip if expected conditions matched those depicted. Both indicators of visitation likelihood were also positively correlated with visitors' overall acceptability rating for the proposed adaptations, although the correlations were weaker.
Only the proposed snocoach (mean D 2.72; sd D 1.41) and helicopter (mean D 2.50; sd D 1.51) adaptations were rated on average as being unacceptable, while the trail (mean D 3.68; sd D 1.31), fence (mean D 3.23; sd D 1.37), and bridge (mean D 3.42; sd D 1.38) adaptations were all deemed acceptable. Given that only adaptations involving motorized transport were rated as unacceptable, we combined these ratings and compared this average to the average acceptability rating for the trail, fence and bridge adaptations.
Results of a paired sample t-test indicated that the snocoach and helicopter adaptations were rated significantly lower, and that the difference was approaching a large effect size (t D 9.74, p < 0.001, d D 0.72).
5. Discussion 5.1. Linking onsite experiences with climate change impacts in parks and protected areas to climate change education and engagement There are many benefits to visiting parks and PAs, including supporting health and wellbeing outcomes, fostering environmental stewardship, and providing a source of revenue for PA management and planning (Eagles 2014;Lemieux et al. 2016). A recent report from the Parks Canada (2014b) recognizes that to realize these benefits, a rich and longterm relationship between citizens and their parks must be fostered through sustainable visitation and use. This report argues that "individuals, communities, employers, and governments at every level and in every department need to see the value of outdoor experiences and their powerful and plentiful benefits" (Parks Canada 2014b, 23). Although true, parks and PAs also face an unprecedented scale of change as a result of direct and indirect climate impacts (Lemieux, Beechey, and Gray 2011a;Lemieux et al. 2011b). If the benefits of experiencing Canada's most iconic landscapes are to be fully realized, individuals, communities, employers and governments must also develop a clearer understanding of how visitor perceptions and consumer behaviour will be influenced by climate change (G€ ossling et al. 2012).
Results here suggest that indirect environmental change will have a significant impact on the pull of the Athabasca Glacier site as a visitor destination. Compared to current conditions, surveyed visitors exhibited a significantly lower preference for the scenario depicting climate impacts. They also rated this scenario as being significantly less natural. Most importantly, potential changes to the site greatly influenced the perceived Note: The 'make current visit' variable had four response categories and was analyzed using non-parametric procedures. a Spearman's correlation. b Pearson correlation. Ã Correlation is significant at the 0.05 level. ÃÃ Correlation is significant at the 0.01 level.

ÃÃÃ
Correlation is significant at the 0.001 level. desirability of visitation, as 22.9% of visitors indicated that they would likely not have made their current trip if they expected to experience conditions similar to the impacts scenario. The implications from a consumer behaviour standpoint are even more severe, given that the vast majority of participants (83.3%), and in particular international travellers, seemed willing to substitute other destinations to engage in glacier viewing.
The willingness to seek glacier viewing experiences elsewhere may be linked to general and site-specific perceptions of climate change. Similar to Brownlee et al. (2013), visitors were quite certain that climate change is occurring and that human behaviour is a key driver. Visitors were also concerned about climate change and largely agreed that the Athabasca Glacier would disappear as a result of future warming. Current educational materials at the site seem to be shaping this latter perception. We only conducted a postexperience survey. However, on average, visitors estimated that the glacier would disappear somewhere close to the year 2100 (median: 80 years). This estimate is in line with Clarke et al.'s (2015) projections of glacial loss (i.e., a 95% reduction in area and volume by 2100) and is almost certainly related to existing educational plaques on site that note the glacier could be gone within a century.
Parks Canada's options to respond to environmental change at the Athabasca site are limited, given the scale and pace of expected glacial decline (Clarke et al. 2015). Clearly, there are few silver linings to the loss of such an iconic landscape. However, the disappearance of North America's most visited glacier does present a bittersweet opportunity to leverage a highly visible climate change impact into a more fulsome public dialogue about the scientific, political, and social realities of climate change in Canada. In support of this dialogue, interactive art installations on site could use visualized landscape scenarios, like those produced here, to present plausible climate futurescapes in a compelling and emotional manner. Adopting an ecosystem approach, installations could incorporate personally compelling narratives that link tangible onsite experiences to more abstract drivers of climate change at broader ecological (e.g., watershed) and political (e.g., national) scales. A similar project was implemented in Marin County, California to depict community impacts of a one-meter rise in sea level (Merchant 2015), and is effectively capitalizing on the less is more (i.e., local, emotional, social, supportive) communicative capacity of visualizations to inform citizens about climate change. Importantly, recent research has shown that enriched educational experiences like this can increase awareness of climate impacts among PA visitors, further supporting the potential efficacy of deploying climate futurescapes in a PA context (Brownlee et al. 2013).

Avoiding maladaptation in parks and protected areas by engaging visitors in adaptation planning
Results of this study suggest that climate change impacts can have a considerable influence on pushÀpull dynamics that shape motivations to visit PAs. Potential climate impacts at the Athabasca Glacier site were visualized and shown to onsite visitors. These visitors rated the likelihood of return visitation significantly lower for scenes showing potential climate impacts compared to scenes showing current conditions. Nearly 23% of visitors also indicated that they would probably not have made their current trip if conditions at the Athabasca site were similar to those shown in the scenes depicting climate impacts. The absolute value of this potential decline in visitation is in line with previous case studies examining visitorÀglacier interactions (e.g., Yuan et al. 2006). However, the implied sensitivity of consumer behaviour to climate-driven environmental change is arguably much higher in our study. The basis for this argument is the fact that visitors in our study responded to visual scenarios depicting only partial glacier retreat, while Yuan et al. (2006) had visitors respond to a verbal scenario where the Yulong glacier had completely disappeared. The higher sensitivity of consumer behaviour to future environmental change implied in our study might be explained by differences in the scale at which visitation was evaluated. Yuan et al. (2006) assessed visitation at a regional scale where motivations were related to amenities other than the glacier. In contrast, we assessed visitation at a landscape scale where there were fewer motivational pull factors. This difference may have made visitors in our study more sensitive to a decline in the glacier itself. Given the very different destination contexts, it is also possible that cross-cultural differences in landscape perception shaped visitor responses (Masuda et al. 2008).
While both of these potential explanations warrant additional investigation, past environmental visualization research suggests that the stronger reaction of visitors in our study likely stemmed in large part from the disclosure contained in the visualized scenes (Sheppard et al. 2008). In other words, seeing future conditions at the glacier made it easier for visitors to evaluate how environmental change might influence their decision to visit or not. If this is the case, findings here indicate that studies that do not employ realistic, visualized scenarios may underestimate the sensitivity of visitation to climatedriven environmental change.
In addition to viewing scenes depicting climate change impacts, visitors responded to scenes that incorporated potential adaptations. These adaptations were designed to reflect best practice from the tourism and climate change literature, as well as direct input from tourism and climate change experts. We expected that scenes showing how adaptations could maintain visitor access to the glacier would lead to more favourable return visitation ratings compared to scenes showing only impacts. Instead, proposed adaptations contributed to a further decline in the stated likelihood of return visitation. Compared to the impacts scenario, a notably larger proportion of visitors (40.7%) also indicated that they would likely not have made their current visit if the conditions at the site included both potential impacts and potential adaptations.
This finding highlights important management considerations linking consumer behaviour and adaptation planning in PAs. First, while large-scale quantitative studies have demonstrated potential growth in PA visitation due to warming (Fisichelli et al. 2015), results here illustrate why case study research is needed to understand how this general trend will play out on a region-by-region (if not park-by-park) basis. Second, the difficulty we experienced in identifying an adaptation pathway that fosters desirable visitor experiences shows why robust stakeholder engagement is critical to effective adaptation planning in PAs. In our case, despite an attempt to design adaptations that fit the character of the site, our expert-defined adaptations underestimated a key dimension of what connotes a desirable experience at the Athabasca Glacier (i.e., naturalness). If practical exercises in designing and implementing adaptation policies are to avoid a similar fate, we recommend PA managers pay careful attention to the fit between the unique identity of protected landscapes and visitors' desire for a natural experience. Not only is this likely to contribute to adaptations that are more sensitive to factors that motivate visitation, it may also reinforce the mandate of PA agencies to prioritize ecological integrity over competing development pressures.
The importance of fit between landscape identity and visitor motivations asks PA managers to broadly consider the social and cultural values that are supported (or not supported) through adaptation. In the specific context of the Athabasca Glacier site and JNP, the notion of fit also questions the foresight of thinking that led to the recent development of the Glacier Skywalk, and that underpins a proposed development at Maligne Lake (both in JNP). In both cases, these development projects reflect an assumption that investment in intensive visitor infrastructure is needed to bolster longterm visitation to Canada's parks and PAs. The negative reaction of Athabasca site visitors to the use of snocoach and helicopter tours documented here calls this assumption into question. Various forms of mechanized and/or large-scale visitation infrastructure do not seem in line with the National Parks Act's mandate to make ecological integrity a first priority in the management of parks and PAs. However, our results suggest that such development projects may also be an affront to the desire for a natural experience that motivates an important segment of PA visitation. At very least, there is reason to question whether developments that threaten ecological integrity are contributing to rich and longterm relationships with our parks, realizing their "powerful and plentiful benefits" (Parks Canada 2014b, 23), or are promoting the idea that our most iconic landscapes are a product for private consumption.

Conclusion
This study paired Dann's (1977) pushÀpull model of visitor motivations with photorealistic landscape representations (i.e., climate futurescapes) to assess the potential impacts of climate change on consumer behaviour in JNP, Canada. Results indicate that 22.9% of visitors would likely not have made their current trip if conditions at the site were similar to those depicted in a 2050 do-nothing climate change scenario. Potential adaptations designed to maintain continued safe access to the glacier were developed with input from tourism experts and were incorporated into a 2050 impacts with adaptations scenario. Under this scenario, 40.7% of visitors indicated they would likely not have made their current trip given the conditions shown. Expert-identified adaptations exacerbated the negative impacts of climate-driven environmental change on consumer behaviour because they did not adequately account for visitor's desire to experience a pristine, natural site.
To our knowledge, this study is among the first to use climate futurescapes to understand the link between climate-driven environmental change, consumer behaviour, and adaptation planning. It provides early and important evidence that supports calls for a better understanding of the link between climate change, visitor perceptions and travel motivations (G€ ossling et al. 2012;Scott, Simpson, and Sim 2012). Findings highlight the potential for expert-driven adaptation programmes to promote maladaptation if they do not adequately account for visitor perceptions of environmental change. Results also call into question the wisdom of recent development decisions in JNP, which could threaten the pull effect of the park's natural identity.
Our findings are in line with previous climate futurescape research (Cohen et al. 2012;Dockerty et al. 2006;Sheppard et al. 2011), but should be considered in light of several limitations. First, to maintain the methodological requirement of measuring multiple perceptual indicators and multiple viewpoints with our onsite participants, we included only two future scenarios. This was done to limit response fatigue. While research from Clarke et al. (2015) suggests that this was sufficient to capture possible physical outcomes at a 2050 timescale, future research could explore a wider range of scenarios by adopting a controlled experimental methodology. Second, resource and access constraints limited our sample to English-speaking participants. Our results cover a wide range of national and international visitors, but a contingent of non-English-speaking tour bus groups who visit the Athabasca Glacier site was not surveyed. As experience suggests that these visitors engage in different visitation activities (e.g., passive viewing from the Columbia Icefield Glacier Discovery Centre), future cross-cultural research should explore whether this group exhibits a distinct set of perceptions regarding climate change and environmental change.
We are encouraged about the capacity of climate futurescapes to contribute to research on climate change and consumer behaviour within the tourism and PA sectors. However, our enthusiasm is couched within broader considerations about the use of environmental visualization technology. The software and hardware used to create climate futurescapes have made astounding advancements in the past two decades. Despite this, recent research suggests that technological sophistication does not guarantee effective communication, and poorly prepared visualization scenarios can significantly bias environmental perception, even when highly realistic imagery is used (Downes and Lange 2015;Lewis 2012;Lewis, Casello, and Groulx 2012). Nassauer (2015) suggests that "visualization is not only powerful as a medium for exchange among stakeholders and scientists, it is insidious for its assumed verisimilitude with real landscape experiences" (2). Powerful environmental visualization technology will continue to shape communication and engagement on the issue of climate change. As it does, we remind visualization preparers and end-users alike that the potential to promote deeper engagement and collaboration is not merely a function of the technology's inherent capacities, but also the agendas and motivations of those who wield it.