A systematic review of participatory scenario planning to envision mountain social-ecological systems futures

Mountain social-ecological systems (MtSES) provide crucial ecosystem services to over half of humanity. However, populations living in these highly varied regions are now confronted by global change. It is critical that they are able to anticipate change to strategically manage resources and avoid potential conflict. Yet, planning for sustainable, equitable transitions for the future is a daunting task, considering the range of uncertainties and the unique character of MtSES. Participatory scenario planning (PSP) can help MtSES communities by critically reflecting on a wider array of innovative pathways for adaptive transformation. Although the design of effective approaches has been widely discussed, how PSP has been employed in MtSES has yet to be examined. Here, we present the first systematic global review of singleand multiscalar, multisectoral PSP undertaken in MtSES, in which we characterize the process, identify strengths and gaps, and suggest effective ways to apply PSP in MtSES. We used a nine-step process to help guide the analysis of 42 studies from 1989 screened articles. Our results indicate a steady increase in relevant studies since 2006, with 43% published between 2015 and 2017. These studies encompass 39 countries, with over 50% in Europe. PSP in MtSES is used predominantly to build cooperation, social learning, collaboration, and decision support, yet meeting these objectives is hindered by insufficient engagement with intended end users. MtSES PSP has focused largely on envisioning themes of governance, economy, land use change, and biodiversity, but has overlooked themes such as gender equality, public health, and sanitation. There are many avenues to expand and improve PSP in MtSES: to other regions, sectors, across a greater diversity of stakeholders, and with a specific focus on MtSES paradoxes. Communicating uncertainty, monitoring and evaluating impacts, and engendering more comparative approaches can further increase the utility of PSP for addressing MtSES challenges, with lessons for other complex social-ecological systems.


INTRODUCTION
Mountain social-ecological systems (MtSES; Box 1) encompass approximately 30.5% of all land (Karagulle et al. 2017, Sayre et al. 2018) and 23% of the Earth's total forest cover (Körner and Ohsawa 2006). Characterized by high levels of biodiversity and endemism, they support an estimated 85% of the world's amphibian, bird, and mammal species (Rahbek et al. 2019a). Based on 2017 population data, MtSES are inhabited by up to 28.3% (2.21 billion) of the global human population (Karagulle et al. 2017, Rose et al. 2018), many of whom are among the world's poorest people (Körner et al. 2017). One in two rural mountain dwellers faces food insecurity, and they have less access to infrastructure and services compared to lowland populations (FAO 2015, Manuelli et al. 2017, FAO and UNCCD 2019. Many local communities rely on the ecosystem services from MtSES regions, such as timber, natural hazard regulation, and tourism, for their subsistence and livelihoods (Harrison et al. 2010). Beyond these MtSES communities, more than half of humanity relies on the freshwater originating in MtSES (Liniger and Weingartner 1998). Indeed, the world's 10 longest rivers have headwaters in MtSES, e.g., the Yangtze River on the Tibetan Plateau and Congo River in the East African Rift (Encalada et al. 2019).
Central to the well-being and survival of much of humanity, MtSES are confronted by climate change, biodiversity loss, land use conversion, and other long-term social-economic challenges (Cuni-Sanchez et al. 2018, Hagedorn et al. 2019, Klein et al. 2019a, b, Rahbek et al. 2019b, Steger et al. 2020. For instance, the rate of warming is amplified as elevation increases (Hagedorn et al. 2019), resulting in higher exposure to climate change in MtSES than the global average or in lower elevation regions (Pepin et al. 2015, IPCC 2018. Such warming can lead to rising cloud bases or reduced overall cloud immersion, which plays an important ecological role in many tropical montane cloud forests by creating isolated patches of habitat with more fog and mist, high soil moisture and carbon storage, and unique species (Bruijnzeel et al. 2011, Helmer et al. 2019. In higher elevation MtSES, glacial retreat has been rapid, with societal implications (Carey et al. 2017, Nyima andHopping 2019). In recent decades, many MtSES have also experienced destabilizing demographic fluxes caused by seasonal migration, permanent rural exodus, amenity inmigration, aging, restructured market relationships, industrial developments, and the abandonment of economic activities (Glorioso andMoss 2007, Park andPellow 2011). Therefore, understanding the challenges to present and future sustainability of MtSES is critical to plan for potential trajectories of change, not only for the rural communities that have developed in relative isolation and whose survival and cultures depend directly on MtSES resources, but also for the lowland communities that depend on resources flowing from these ecosystems (FAO 2015).
Yet, planning for sustainable, equitable transitions in MtSES is a daunting task considering the range of uncertainties and the peculiar conditions of MtSES, referred to as paradoxes (see Klein et al. 2019b;Box 2). Uncertainties in future planning are associated with a high occurrence of cross-scale interactions in MtSES, such as cascading hydrological changes at watershed/ basin scales ) and often mismatched supply and demand of ecosystem services across elevational gradients (Brunner et al. 2016). Uncertainties in possible futures are highly contingent and nonlinear, and are affected by events and decisions that are themselves often highly unpredictable (IPBES 2016). Planning for transitions is further complicated in that decisions based on a specific prediction can lead to unintended consequences for sustainability and equity, particularly when scientists assessing conservation priorities do not use participatory approaches, have different values than MtSES stakeholders, and these hidden value-judgements are not made explicit (Huber et al. 2013, Seidl 2015. Poorly defined problems related to resource allocation can also arise from a lack of scientific advice, e.g., the Green Revolution in the Himalayas (Rasul 2010, Game et al. 2013).

Box 1: Definition of MtSES.
MtSES are typically characterized by high kinetic energy, steep vertical gradients, and are more likely than nonmountain regions to experience multihazards such as earthquakes, landslides, avalanches, flash floods, erosion, and fire (Gardner and Dekens 2007, Zimmerman and Keiler 2015, Klein et al. 2019b. One recent attempt to characterize global mountain extent is the K3 geospatial raster, which uses three classification parameters (slope, relative relief, and profile) to define four mountain classes of plains, hills, mountains, and tablelands at 250-m resolution (Karagulle et al. 2017, Sayre et al. 2018. However, different definitions of mountains have been proposed for various contexts and countries. This is in part because definitions established for MtSES boundaries are determined not only by geological conditions and landforms but also by climatic and hydrological conditions, ecosystem patterns, animal and plant distributions, and human activities, e.g., mountaineering, mining, and leisure Ohsawa 2006, Körner et al. 2017). Definitions also diverge because the biophysical boundaries of MtSES can be difficult to align with administrative or economic boundaries (Price et al. 2018). In this paper, we use the K3 definition to analyze global mountain distributions in relation to population density, because it is the most recent, accurate, and finest resolution dataset available. However, our concept of MtSES is more holistic than the biophysical K3 definition, encompassing complex adaptive systems consisting of interacting social-ecological processes and unique challenges (Box 2).
Participatory scenario planning (PSP) can help overcome some of these challenges for sustainable and equitable transitions in MtSES by using novel combinations of expertise to creatively envision and critically reflect on a wider array of innovative pathways for adaptive transformation (van der Heijden 1996, van Vliet 2011, Bai et al. 2016). The central idea of PSP is to consider a range of diverse, plausible futures, e.g., events, conditions, and uncertainties, rather than focusing on the prediction of a single or most probable outcome (Wilkinson and Eidenow 2008). Through close collaboration with diverse stakeholders, the process can build consensus, trust, and cooperation among participants, and improve the legitimacy and understanding of scenario outputs , Allington et al. 2018. PSP offers a valuable means to bridge the science-policy interface, by accommodating varied perspectives, needs, expectations, and values, and by coproducing an integrated understanding of landscape and socioeconomic dynamics (Peterson et al. 2003, Wilkinson 2009, IPBES 2016. PSP can also help address some of the challenges created by MtSES paradoxes (Box 2). For example, PSP can increase decision makers' abilities to identify when, where, and why scarcities occur in MtSES, e.g., water availability due to changing snowpack melt . PSP can encourage MtSES adaptation to global change by examining system drivers, identifying the underlying mechanisms most relevant for unpredictable futures, and evaluating which interventions are most appropriate (Soliva andHunziker 2009, Fischler et al. 2016). At the same time, local knowledge that emerges through PSP can help fill critical gaps in understanding the functioning of MtSES when data are scarce (Oteros-Rozas et al. 2015). PSP is an opportunity not only to open local knowledge systems to global trends that may affect isolated and remote MtSES, e.g., through telecoupling, but also introduce local knowledge to those living outside MtSES, mainstream it into decision-making processes (Capitani et al. 2019), and assess trade-offs between key ecosystem services such as water, biodiversity, and carbon (Capitani et al. 2019). PSP may be a better tool for addressing MtSES challenges than other methods, e.g., participatory GIS, integrated assessment models, and ethnographic or economic valuations, because of the explicit attempt to not only help people to understand, prepare, and adapt more effectively to future events, but also to develop actionable pathways to challenge and change these events (Kahane 2012).

Search strategy
A systematic review of the literature on PSP in MtSES followed methods established by the Collaboration for Environmental Evidence Guidelines and Standards for Evidence Synthesis in Environmental Management (CEE 2013). This method is used widely and recognized as a standard for accessing, appraising, and synthesizing scientific information to inform decision making. To minimize bias, improve reporting, and ensure a highquality and comprehensive systematic review, we followed the preferred reporting items for systematic reviews and metaanalyses (PRISMA) checklist (Moher et al. 2009). This checklist corresponds to the iterative stages of our structured review process (Fig. 1); PRISMA helps authors ensure a transparent and complete reporting of their review process by using systematic and explicit methods to identify, select, and critically appraise relevant research, and to collect and analyze data from the studies included in the review. Following these protocols ensured rigor, objectivity, verifiable procedures, and clarity of the study design (CEE 2013). We developed the search strategy for finding MtSES PSP studies through (a) discussions with researchers and stakeholders at a workshop on participatory modeling in MtSES; (b) consultations with an advisory board of scientific experts in PSP in MtSES (Appendix 1) throughout 2017-2018; and (c) working with an experienced environmental librarian.

Sources of publications and key search terms
We conducted systematic searches of peer-reviewed and nonpeerreviewed literature in bibliographic databases, key international journals, specialist organizations, online databases, and search engines, as well as via an open webinar with the extended Mountain Sentinels Collaborative Network and consultations with its advisory board (between 22 April and 8 November 2017). Five bibliographic databases were searched: Thomson Reuter's (formally ISI) Web of Science™, Core Collection Academic Search Premier, CAB Abstracts published by CAB International (1973-present), AGRICOLA National Agricultural Library and Citation Database and Social Sciences Full Text (H. W. Wilson). A repeat search on 4 November 2019 in Web of Science found only three additional studies published since the initial review.
Five key international e-journals whose topic areas closely aligned with the research question were hand searched. Google Scholar was used to retrieve the first 200 search results and we checked for studies not captured in the above databases. Fourteen subjectspecific web sites including nongovernmental organizations, public and research institutions, and online databases were searched for reports, conference proceedings, policy briefs, book chapters, and individual research papers (Table 1). To assist in screening nonpeer-reviewed literature, text extraction software was used to identify key words related to the research questions.  (Pullin et al. 2013;Appendix 3, 4). The final string that resulted in 43 relevant studies when tested in Web of Science consisted of the terms: "scenario analy*," "scenario develop*," "scenario plan*," "scenario," "mountain*," "participat*," "collaborat*." These terms were applied to all databases, subject to their individual search requirements. Boolean operator terms and wildcards, i.e., a character that can be used to substitute for other character(s) in a string, connected search terms, which were disaggregated using truncation ("*" in most databases). The final searches produced a total of 1989 studies for further screening.

Study inclusion criteria
Studies were included in the review if they do the following: report data on PSP with local stakeholders; include stakeholders living permanently/seasonally in MtSES, or stakeholders affected upstream/downstream, e.g., by ecosystem service provisioning or migration; address interactions of social and ecological systems (van Notten et al. 2003, Binder et al. 2013, or adaptive management (Holling and Allen 2002, Walker et al. 2002, Folke 2007; use qualitative or quantitative data, or both; are available in electronic format; are published in English, given the linguistic competency of the review team; and defined their system of focus to include MtSES facing single-and multiscalar, multisectoral challenges. All publications were included that were published up to November 2017. Of the papers that met the inclusion criteria, only three studies were excluded from the analysis because they were published in other languages, i.e., German, French, and Spanish.

Data screening and critical appraisal
Six reviewers conducted screening at title, abstract, and full text stages to assess whether studies met the inclusion criteria listed above and chronicled the reason for exclusion in a master Microsoft Excel database ( Fig. 1, Appendix 5). At title and abstract screening, Randolph's free-marginal Kappa coefficient, i.e., degree of chance-adjusted agreement, on a random subset of 100 studies was 0.72, which falls above the normal agreement range of > 0. 6-0.7 (Randolph 2008), indicating that there was sufficient agreement among reviewers. Where there was doubt, we retained the studies. Ambiguities were discussed in regular meetings, and secondary reviewers verified decisions. We used five quality criteria to assess whether studies were of sufficient quality for inclusion ( Fig. 2; see Rodríguez et al. 2016).
The development of a nine-step process to guide data coding, extraction, and evaluation Forty-two studies met the inclusion and quality criteria. We created a codebook (Appendix 6) that was developed and tested on six case studies by two independent investigators. To help structure our coding (and thus, exploration) of the studies, we used themes taken from a synthesis of key issues for MtSES sustainability (Klein et al. 2019b), the global planetary and social boundaries frameworks (Rockström et al. 2009, Raworth 2012, and a nine-step process for conducting PSP that we developed based on a literature review and expert contribution. The ninestep process consisted of methods that were commonly applied in PSP tools and typologies broadly (both within and beyond MtSES), depending on study context, purpose, and goals (e.g., Scholtz and Tietje 2002, van Notten et al. 2003, Oteros-Rozas et al. 2015Fig. 3). Based on these nine steps, we developed 89 questions that we used to evaluate the extent to which the 42 studies included information about each step. In our evaluation of each study, we also included several open-ended questions pertaining to the primary livelihoods mentioned in each case study, and any strengths and weaknesses of the PSP process. Rather than using a set of pre-existing, theoretically informed codes for these open-ended questions we developed codes inductively. Finally, we used a Likert scale to assess the degree to which each study explicitly addressed uncertainty. Six reviewers evaluated the studies (between 9 November 2017 and 2 February 2018). Each reviewer met with a second reviewer on a weekly basis to discuss and resolve inconsistencies in their coding. See Appendices 7 and 8 for a full list of the case studies analyzed. Each study that met the inclusion criteria (n = 44) received a total quality assessment score of 0-5, based on the sum of its scores for each quality criterion (yes = 1, no = 0). Only studies with total scores of 3-5 were considered eligible for further data extraction.

Analysis
We generated summary statistics based on the frequency of citations and conducted qualitative and quantitative syntheses on trends. We mapped case studies in ESRI ArcGIS 10.5 (ESRI 2015). Point locations of case studies were identified either directly from the reviewed manuscripts, by contacting study authors, or estimated using other geographical information provided in the text. We assessed their geographic distribution relative to the distribution of mountain areas and human populations in MtSES. We defined geographic regions following the United Nations Statistics Division's Intermediate Regions. We calculated mountain areas by region from the 250-m resolution K3 mountain definition (Karagulle et al. 2017; Box 1) and mountain population using the 30-arc-second LandScan™ 2017 dataset (LandScan 2017™, ORNL, UT-Battelle, LLC; Rose et al. 2018).

Fig. 3.
A flow chart of the proportion of studies within the systematic review (shown in dark grey in the pie charts and percentages) that provided information on each step of the nine-step process that we used to analyze participatory scenario planning (PSP) case studies in mountain social-ecological systems (MtSES). Appendix 9 provides additional detail. The process starts by establishing the goal of the PSP and defining the context boundaries. Stakeholders then envision desired and undesired futures, i.e., imagining forward-looking strategies, interventions, or innovations that could inspire transformative change. Following this, the internal and external drivers of change within the defined SES boundaries are identified. Narrative storylines are then constructed, which can be transformed into quantifiable models and tested for consistency and plausibility. Process and results are communicated to relevant stakeholders for dissemination and feedback. The PSP process and outputs are then monitored and evaluated to inform future iterations, policy and management priorities, and research.

Geographic application
Case studies meeting the inclusion criteria were spread across 39 countries and five continents (Fig. 5). PSP in MtSES has primarily been applied in Europe (54.8%, n = 23 studies). Eight studies were conducted in Asia (19.1%), half of which were in China; five in Africa (11.9%), four of which were in Tanzania; four in North America (9.5% ea.), all in the United States; and two in Oceania (4.8%), both in the Australian Alps. Almost half of the PSP studies (45.2%, n = 19) were defined by political or administrative units, while 28.6% (n = 12) were defined by natural features and 21.4% (n = 9) were defined by both. One-third of studies (n = 14) were developed at the landscape scale. In terms of the spatial scale, Ecology and Society 25(3): 6 https://www.ecologyandsociety.org/vol25/iss3/art6/ six studies were conducted at the farm or village/community level (14.3%), four at district level (9.5%), nine at regional level (21.4%), four at national level (9.5%), and five at the international level (11.9%). Although many MtSES ranges span international political borders, e.g., Mount Elgon, Rwenzori, the Pyrenees, Alps, and Himalayas (Körner and Ohsawa 2006), only five studies researched transboundary MtSES in Asia (n = 2) and Europe (n = 3). Eight studies (19.1%) explicitly used a spatially nested, multilevel approach.  (black) is commensurate to the growth of publications that report on scenarios, both within the field of environmental science (*1000, dark grey) and in all fields (*10,000, light grey), as indicated in a Web of Science search over the same period.
Many of the studies conducted PSP at multiple sites within continents (n = 127 sites). When considering this finer spatial resolution, European sites emerge as particularly overrepresented in the literature (63.8%) relative to both the proportion of global mountain area contained in Europe (17.9%) and the proportion of the global mountain population inhabiting European MtSES (8.0%; Fig. 6). East Africa is also overrepresented (13.4%) relative to its proportion of mountain area (5.8%), but less so when considering its share of the global mountain population (9.6%). No sites were located in mountain regions of Northern, Central, and Western Africa, Central and Western Asia, or Central and South America, despite South America containing a relatively high proportion of the global mountain area (12.3%) and population (7.7%). North America is also under-represented relative to its proportion of mountain area (13.1%), although its share of the global mountain population is low (2.2%). East Asia is most severely under-represented (3.9%), considering that it contains the largest proportion of mountain area (17.3%) and highest population of any region (29.6%).

Temporal scope
The mean time span between the scenario creation and the projected year was 33 years, but scenarios ranged from 14 to 90 years into the future. The mean baseline year for scenarios was 2009, ranging from 2000 to 2015, while the mean target year for scenarios was 2042, ranging from 2020 to 2100. Thirty-eight studies (90.5%) used a single target year. Only eight studies (19%) used a midterm horizon year to mark incremental progress toward final scenarios, with a mean of 2023, ranging from 2010 to 2040.

Fig. 5.
Map of study locations. The inset map depicts the European region. Some of the study areas indicated with black dots closely overlap. The blue shaded areas represent mountains, as defined by Karagulle et al. (2017). Definitions of mountains by the authors of the studies included in this review may differ.

The nine-step PSP process
We explored the extent to which studies employed each step in the nine-step PSP process (Figs. 3 and 7), as described below. We found 50% of the studies addressed at least eight of the nine steps outlined. Fig. 6. The distribution of participatory scenario planning (PSP) study sites differs among geographic regions. The percentage of PSP study sites in each region (relative to all PSP study sites) is compared to the percentage of mountain area located in each region (relative to the total global mountain area; (a) and to the percentage of people living in mountains in each region (relative to the total global mountain population; (b). In (a, b), points with numeric labels are coded as follows: 1 = Western Africa, 2 = Central Asia, 3 = Northern Africa, 4 = Central Africa, 5 = Western Asia, 6 = Caribbean and Central America. Dashed lines represent a 1:1 relationship. Population density in mountains varies across geographic regions (c). Geographic regions are drawn from the United Nations' definition of Intermediate Regions (d).
Step 1. Establish PSP goal Most studies (97.6%, n = 41) stated a clear PSP goal at the beginning of their project. The most common goal (or purpose) was to improve understanding among stakeholders (78.6%, n = 33), by explicitly integrating their diverse views into shared representations of the future. Decision support was the goal of 45.2% (n = 19) of studies, often focusing on ecosystem service management, REDD+ initiatives, climate change mitigation to reduce greenhouse gas emissions, or national planning. The next most frequently stated goal of PSP was for prediction (31.0%, n = 13), followed by learning (11.9%, n = 5) and communication (9.5%, n = 4). Building on categories created by van Notten et al. (2003), we determined that the goal of PSP was exploratory (i.e., creating scenarios to examine plausible drivers of change) in 61.9% of studies (n = 26), prepolicy (i.e., creating scenarios to examine futures according to their desirability) in 23.8% of studies (n = 10), and both exploratory and prepolicy in 14.3% of studies (n = 6).

Fig. 7.
The percentage of studies that employed each step of the nine-step process for participatory scenario planning (PSP) in mountain social-ecological systems (MtSES). The themes (2a-d) are defined when determining boundary conditions. These form the main focus of alternative futures explored in the scenarios.
Step 2. Define social-ecological system boundaries All studies defined the system boundaries prior to developing scenarios. These boundaries entail multiple aspects, such as the climate, biomes, land use, and livelihoods of the MtSES and features of the PSP process, such as the thematic focus, number or diversity of stakeholders involved, and the length of their involvement. Recognizing that these categories are nonexclusive, we find that most studies were defined by issues (73.8%, n = 31), more so than by geographic areas (14.3%, n = 6) or the types of stakeholders or institutions involved (7.1%, n = 3).
The most prominent stakeholders participating in the processes were district, regional, and national government officials (57.1%, n = 24), land managers (52.3%, n = 22), and conservation groups, park authorities, and NGOs (47.6%, n = 20). Municipal institutions, community councils, or tribal and indigenous organizations were only represented in a third of cases (33.3%, n = 14), followed closely by private (31%, n = 13) and academic (21.4%, n = 9) institutions. No studies included bilateral or multilateral organizations. The number of participants ranged from 11 to 240 individuals, averaging 48.4 ± 9.9 (mean ± SE). PSP studies most frequently had three different types of stakeholder groups participating in workshops (23.8%, n = 10), although this ranged from 1 to 10. Few included more than six types of stakeholders (9.5%, n = 4). Stakeholder engagement typically lasted one year (47.6%, n = 20), followed by 1-4 years (23.8%, n = 10). Only one study indicated long-term (> 10 years) engagement with stakeholders in the process. The PSP process was described as embedded in a larger research program in 40.5% (n = 17) of cases. (See "Thematic coverage" above and Fig. 7 for additional details on the use of themes.) Step 3. Envision desired/undesired futures All reviewed MtSES PSP studies included envisioned desired and undesired futures. Less than half of the studies discussed synergies and trade-offs (40.5%, n = 17) between desired and undesired futures.
Step 5. Construct scenario storylines Most scenarios included storylines or narratives, i.e., qualitative description of future developments (90.5%, n = 38). Scenarios were normally constructed using forecasting (90.5%, n = 38) rather than backcasting (4.8%, n = 2), while two studies used both (4.8%). Data were typically collected from workshops (64.3%, n = 27), desk-based research (50%, n = 21), interviews (40.5%, n = 17), and focus group discussions (23.8%, n = 10). Twenty-six studies used a combination of two to four of these methods to construct the scenarios, whereas only two studies used a single method. On average, two to three workshops were held over the PSP process (ranging from zero to eight). Over half used only local-level scenarios (57.1%, n = 24), while fewer used only regional/global scenarios (11.9%, n = 5), and about one-third of studies used a hybrid of local and regional/global scenarios (31%, n = 13).
Step 7. Consistency and plausibility analysis Only 9.5% of studies employed both consistency and plausibility analyses (n = 4). Very few (19%, n = 8) conducted a test for consistency, whether internal, i.e., reviewing whether impact variables within a narrative can occur in combination, or external, i.e., whether diverse future states, or local/global scenarios contradict one another. One-third of studies (33.3%, n = 14) conducted a test for plausibility, i.e., if the scenario falls within the limits of what might conceivably happen. Methods included historical or expert validation, comparing findings to published results from neighboring areas and/or other models, and assessing policies, plans, and actions to identify barriers and bridges to desired outcomes.
Although divergent futures implicitly capture uncertainty, only in 18 cases did the six reviewers strongly agree (7.1%, n = 3) or agree (35.7%, n = 15) that studies explicitly addressed uncertainty. Three main methods were used in the studies to address uncertainty: positioning scenarios along two axes representing extremes (e.g., high/low), ranking scenarios using a scale, or discussing limitations and data resolution. No studies measured stakeholders' degree of confidence in the data inputs or outputs.

Step 8. Cocommunication of PSP process and results
Dissemination of results to stakeholders involved in the PSP process was mentioned in half the studies (50%, n = 21). Dissemination aimed to summarize findings, ensure results are understood, obtain feedback, and discuss intentions to apply the evidence to real-world challenges. In most cases, the target audiences were participants who developed the scenarios (45.2%, n = 19) or researchers (42.9%, n = 18). Occasionally, intended audiences were decision makers operating at sub/national levels. To communicate results beyond the group involved in the PSP process, the main outreach materials were scientific publications (85.7%, n = 36) and reports (28.6%, n = 12). The main tools were knowledge representation or cognitive conceptualization diagrams, which represent system entity, processes, and https://www.ecologyandsociety.org/vol25/iss3/art6/ interactions (35.7%, n = 15), spatial representation tools (33.3%, n = 14), narratives (16.7%, n = 7), or other simulation tools (16.7%, n = 7). Often a combination of tools, such as maps, diagrams, and three-dimensional landscape visualizations, was used to help different audiences, such as local inhabitants and authorities, to conceptualize models (40.5%, n = 17).
Step 9. Monitor and evaluate process and outcomes No studies conducted both monitoring and evaluation. One study monitored outcomes over one year using systematic data collection to track progress using indicators . Most studies (66.7%, n = 28) did not formally evaluate the design, implementation, results, or consequences of social learning. Half of all studies included in the review (n = 21) provided evidence of short-and long-term outcomes, while few (7.1%, n = 3) used outputs to inform further research. Evaluation methods included qualitative, self-reflexive assessments in focus groups, interviews, surveys, literature reviews, participant observation, and expert meetings. Table 2 summarizes the strengths and weaknesses of the PSP process reported by participants.

DISCUSSION
Our evaluation of how PSP has been employed in MtSES, including a characterization of the process, is organized around the following key areas. First, we evaluate the geographic, temporal, and thematic foci of the studies. We then examine the benefits of PSP for MtSES using the nine-step PSP process, explicitly considering the goals of the process (Step 1), the stakeholders/institutions involved and how the boundaries of the MtSES were defined (Step 2), the identification of drivers of change (Step 4), the cocommunication of the PSP process and results (Step 8), and the evaluation of process and outcomes (Step 9). Steps 3 and 5 are examined in detail in a subsequent paper. Finally, to identify key gaps in PSP for MtSES, and what can be learned from the application of PSP in other SES, we analyze which steps in the nine-step process were frequently omitted.

Geographic application
Although MtSES research is on the rise (Gleeson et al. 2016), the geographic coverage of PSP in MtSES remains biased toward European countries, with less focus in Asian, African, and Latin American MtSES, relative to their total extent and total human population. Globally, population density is highest in Western and Southern European MtSES (Fig. 5), where 32.3% of PSP sites are located. By contrast, East Asian MtSES cover the largest area and host the highest total human population of any MtSES globally, but only 3.9% of PSP sites are located there. South America has a relatively large share of mountain area (12.3%), but completely lacks PSP studies. Although these results may be an artefact of our language selection criteria, we consulted with MtSES experts working in these regions, who confirmed that to their knowledge no published records of PSP exist to date (although other methods to assess SES may have been applied). This finding reflects similar biases in broader conservation research, where comparatively less research is undertaken in the world's most biodiverse countries (Meijaard et al. 2015), and the science conducted is often not led by researchers based in-country (Boakes et al. 2010, Thorn et al. 2016, Wilson et al. 2016. A lack of MtSES PSP application in developing countries may bias any synthesis effort toward the values, needs, and context of regions where it is more frequently conducted, i.e., Europe, with the risk that these conclusions may then be translated incorrectly to other regions (Hovland 2003). Therefore, commonalities and divergences in trajectories of change across MtSES should be analyzed by taking into account SES across an array of socioeconomic-cultural conditions. Efforts to synthesize across global MtSES must remain cognizant of these biases, and work to mitigate them. For example, the MtSES research community needs to closely consider ways to address the entrenched disadvantages some countries have when initiating research projects, e.g., access to funding, in-house research support offices, language (Meijaard et al. 2015). There is also a need to reduce the logistical and financial burden of studying across international political borders. In particular, regional organizations have a key role to play in enhancing transboundary cooperation, as evidenced by the PSP work in the Hindu Kush Himalaya by the International Centre for Integrated Mountain Development (Roy et al. 2019).
Although it has been suggested that MtSES are best analyzed across scales, where material and social connections between the highlands and lowlands become clear ), our review revealed few scenarios with this type of nested, multilevel approach. This is not surprising because multilevel analyses are more complicated and require a greater diversity of actors and resources. The spatial scale selected could reflect the priorities of actors responsible for managing adaptation pathways or of institutions driving the PSP process. More multiscale spatial and organizational integration is therefore required across the different phases of PSP to improve the extent to which different MtSES perspectives are incorporated , Biggs et al. 2007, Mistry et al. 2014, Rosa et al. 2017.

Temporal scope
There has been a steady increase in studies concerning MtSES since 2006, perhaps mirroring the growth in international global climate and biodiversity assessments that have increasingly referenced PSP research, e.g., Global Biodiversity Outlook and subglobal Millennium Ecosystem Assessment scenarios. For instance, spikes in 2013 and 2015 occurred when planning processes increasingly adopted PSP to envision future pathways and social science methods, e.g., International Panel on Climate Change (IPCC) Fourth Assessment or Sustainable Development Summit, and when the International Panel on Biodiversity and Ecosystem Services (IPBES) was launched. These trends indicate a growing interest in MtSES futures, both locally and meeting large-scale sustainability challenges, whilst demonstrating increased investment in MtSES research.
The temporal scope of PSP in MtSES is wide ranging. Many studies selected medium-term time horizons (~30 years), which may be associated with forecasts of human population or species distributions, successional change in forests, the generational memory of land managers, infrastructural lock-in, or multidecadal climate change projections. Near-term horizons of 14 years, the shortest of any study in this review, may be associated with municipal planning or business cycles, institutional memory, or international strategies such as the Sustainable Development Goals. The longest time horizon of scenarios was 90 years, Table 2. Strengths and weaknesses reported by stakeholders participating in the studies, showing that there are clear benefits to participatory scenario planning (PSP), e.g., consensus building and dialogue, illuminating assumptions and interconnections, but also limits to the process and use of PSP, e.g., extent of representativeness for concrete decision making, knowledge-and resource-intensive. SES = social-ecological system. Strengths Allington et al. 2018 appropriate to envision structural changes in the 21st century, such as transitions from subsidized, centralized, and large-scale reliance on fossil fuels to greater energy efficiency, lower energy consumption, and sustainable building designs . Capturing this range of scales is important in MtSES, considering that cycles of resource abundance and scarcity may occur at different repeat intervals, e.g., interannual or decadal drought cycles and intra-annual seasonal variability, and therefore long-term planning and program longevity are needed . Furthermore, exploration of nested temporal scales is needed because people's perceptions of the short-term is often more certain, relevant to current needs, and easy to envision compared to the long term. This is analogous to spatial scales, where connections across smaller landscapes are clearer than connections linking MtSES with lowland SES (Huber et al. 2013), or when telecoupling links distant markets in complex social-hydrological relationships (Chignell and Laituri 2018).

Thematic coverage
Key themes that emerge as important for the future sustainability of MtSES are governance, economy, land use change, and biodiversity (Fig. 7, Appendix 9). These themes, which we describe below, resonate with those raised in other place-based PSP literature (e.g., Oteros-Rosaz et al. 2015). There are key themes that were not addressed in our review, despite their importance in MtSES literature. Planetary boundaries (Rockström et al. 2009) such as stratospheric ozone depletion and nutrient cycling were not addressed, likely because they lie beyond the immediate concerns of MtSES stakeholders. Two themes from the social boundaries (Raworth 2012) that we would expect to be important in MtSES, gender equality, e.g., women's engagement in upland agricultural markets, and adequate sanitation for all, e.g., open defecation-free status, have also been overlooked , Molden et al. 2014, Budhakoti et al. 2017, Klein et al. 2019b. Similarly, the repercussions and opportunities of permafrost thaw (Yang et al. 2018, Nyima and Hopping 2019) https://www.ecologyandsociety.org/vol25/iss3/art6/ and use and trade of medicinal resources, including high value commodities Castella 2011, Hopping et al. 2018), received little attention.
Governance: Our analysis suggests that MtSES would benefit from improved horizontal (across space) and vertical (across levels of organizations) governance, and responsive institutions, alongside improved communication, knowledge coproduction, and cooperation among key actors, especially when there are contested views . Securing the livelihoods of MtSES communities rests on better inclusion of local landowners in decision-making processes and action at the subnational level. Enhancing transboundary cooperation across political borders, particularly in relation to transportation infrastructure, may facilitate domestic and international accessibility to remote MtSES regions and thereby improve market access, infrastructural development, and tourism. Strong social safety nets, effective local leadership, informal networks, and strong national institutions combined with international mechanisms will be needed so that marginal upland communities, who often manage MtSES commons, have secure resource rights .
Economy: Our review highlights the need for greater attention to trade-offs and equity in MtSES economic transitions. Scenarios identified the potential decline or replacement of traditional sectors, e.g., pastoral, agricultural, and timber, by industrial and service sectors, e.g., tourism, settlement, and manufacturing ), which could lead to depopulation, land abandonment, and the loss of traditional practices and cultural heritage in upland areas . Conversely, this could encourage changing patterns of in-migration and reruralization as a result of new employment opportunities, investments in upland areas, lifestyle changes, and new market opportunities for niche products . Increased tourism may also significantly benefit communities in MtSES if revenues are used for species conservation and providing public services, but could also lead to cultural knowledge loss and pollution . Emerging opportunities lie in payments for ecosystem service schemes including carbon markets, reforestation, restoration, afforestation, sustainable charcoal production ), but with shortcomings, such as windfall profits or elite capture. Many scenarios foresee MtSES farmers shifting away from subsistence agriculture to cultivating for recreational purposes, i.e., leisure farming and eco-/agri-tourism (Enfors et al. 2008, Soliva and. Improvements in telecommunication, infrastructure, information communication technology, telebanking, and access to remote sensing technology could lead to more off-farm livelihood diversification and specialized market chains , Daconto and Sherpa 2010. However, the extent to which MtSES will benefit from technological innovation is contingent on a wide spectrum of legal, governance, economic, institutional, and environmental factors (Capitani et al. 2016, Roy et al. 2019).
Land use change: In the future, landscape multifunctionality, connectivity, land tenure rights, and spatial planning, including land demarcation and zoning , will play an increasingly important role for sustainable management of MtSES. Our analysis suggests that changing demographic and migration patterns will cause a shift in land uses and their associated ecosystem services, resulting in new sustainability challenges in MtSES. For example, food insecurity may become an even larger challenge if people move away from traditional agricultural-based production systems toward more industrialized systems , calling for more multifunctional landscapes. Investment support from government, private sector, and local associations can improve the future viability of agricultural and pastoral activities in MtSES , particularly in areas under threat from imminent land privatization, land scarcity, and climate change (Allington et al. 2018. In this regard, strengthening property rights could provide a mechanism to secure communities' long-term survival, by buttressing against private investors moving into upland areas, mitigating top-down control by governments, and benefiting more widely from new financial streams for conservation , Daconto and Sherpa 2010.
Biodiversity and water provisioning services: Rates of biodiversity loss are expected to continue, both in MtSES and elsewhere. Drivers of biodiversity loss in MtSES include expanding human populations, poor protected area enforcement, and growing energy demands . For highelevation species, their distribution and community composition will likely change, with northward/uphill migration where habitat is available, establishment of new and/or invasive species, and species adaptation or extinction , Körner et al. 2017. There is also the risk of declining genetic diversity in MtSES because of competition with invasive or exotic species, e.g., high-value eucalyptus replacing conifers, and the introduction of genetically modified organisms . MtSES communities will have to prepare for a world with scarce freshwater resources, especially in light of growing water demands and contamination from agriculture, mining, and manufacturing , UDSM IRA et al. 2016, and altered ground and surface water availability and hydrological flows across elevations with climate change .

Increased learning
Although learning was rarely explicitly listed as a goal of the PSP process, there is unmet potential for learning to be a valuable outcome. In our evaluation of process and outcomes (Step 9), studies indicated that participants learned by integrating diverse views, understanding the local context, and comparing trajectories of change in and across MtSES. PSP improves social learning (Keen et al. 2005, Reed et al. 2010) and systems thinking (Dyball et al. 2007, Keen andMahanty 2006), allowing stakeholders to better understand complexities that exist in MtSES. Many of these processes can be used to foster strategic foresight, encouraging learning from past events , Seidl 2015. The PSP process can thus be as important as the outcome. Yet, papers rarely present post-hoc evidence to demonstrate this learning, and the resulting benefits are generally inferred rather than clearly evidenced. We identify the need for more follow-up research to discover whether participants perceive lasting benefits from the PSP process well after it has concluded.

The inclusion of diverse actors in decision making
PSP in MtSES is being used predominantly as a tool to build cooperation and collaboration, shared understanding, and decision support (Step 1). However, there is often low participation in terms of diversity and number of stakeholders at multiple stages. For example, bi-and multilateral institutions have rarely been included in PSP in MtSES, despite their role in issues such as coordination of international environmental agreements, e.g., in the Himalayas (Daconto and Sherpa 2010). Similarly, less than half of the studies define their study area using political or administrative boundaries (Step 2), even though doing so would potentially make scenarios more easily translatable for policy decisions. Remarkably, studies rarely engage nonacademics to define system boundaries and develop questions to be addressed (Step 2), even though these actors are often the intended end-users of the results. By not doing so, PSP may bias patterns of international cooperation and budget allocation, leading to unintended consequences for MtSES.
There are a variety of approaches to identify stakeholders for inclusion in PSP research, including stakeholder analysis and actor mapping, based on different ways of characterizing stakeholders and their relationships to other stakeholders (Reed et al. 2009). Developing stakeholder proficiency in envisioning can be a time-and engagement-intensive process , particularly for isolated MtSES communities who may find it difficult to imagine divergent futures from present/usual conditions because having to select a limited number of drivers of future change might contradict participants' lived experiences in multifunctional MtSES (FAO 2015). Yet, local knowledge can provide critical insights into MtSES processes (e.g., Klein et al. 2014, Thorn 2019, Steger et al. 2020, and the inclusion of diverse actors in PSP is essential for understanding and managing appropriate future trajectories of MtSES that fulfill the needs and desires of local communities.

Research collaborations
Our research indicates that there is currently a limited set of institutions and researchers employing PSP in MtSES. Teambased scientific work is common in MtSES PSP, with 50% of studies being coauthored by some of the same researchers. Such collaborations could accelerate innovation. However, the limited set of institutions employing PSP in MtSES could indicate geographic and disciplinary biases, influencing how questions are framed, and which variables are assessed, which ultimately affects the outcome of studies (Step 4;Friedman et al. 2018). Similarly, authors' origins or institutional affiliations could impact the locally or regionally relevant understandings of unique MtSES contexts (Karlsson et al. 2007, Wilson et al. 2016. We encourage long-term capacity building in PSP, especially of early career researchers involving both South-North and South-South exchange (Gleeson et al. 2016).

Visibility of scholarship
Half of the MtSES PSP studies are distributed online without restrictions on use and reuse (Step 8), which is comparable to open-access publishing trends for scientific studies in general (Archambault et al. 2014). Increasing the rate at which MtSES PSP studies are made freely available online would be particularly valuable because of the relevance of PSP research to MtSES practitioners (Oteros-Rozas et al. 2015). Publicly available tools and insights from PSP could enhance knowledge democratization and improve the visibility and impact of PSP scholarship in national, e.g., public safety and risk reduction agencies, and international arenas, e.g., IPCC (Wilson et al. 2016). There are economic models of open access publishing that avoid author processing charges for publishing fees, which is perhaps the primary barrier to open access publishing, although other barriers exist, such as internet access. Knowledge products and methods of dissemination to inform local decision makers, e.g., policy briefs and meetings, can be strengthened alongside open access publishing. In most cases, the target audiences were participants or researchers who developed the scenarios, therefore potentially excluding influential decision makers who lacked the time to commit to these processes. This trend could limit how research outcomes shape stakeholders' ability to take up research and engage in agenda setting. Although it is possible that studies did not necessarily report all their communication outreach, horizontal learning opportunities and nonacademic venues for disseminating research could hasten broader sharing and use of PSP, thereby providing the public with a valuable tool for anticipating global environmental change (Hovland 2003).

What are key gaps of PSP in MtSES, and what can be learned from PSP in other contexts?
Results from our evaluation point to two key gaps in PSP for MtSES: communicating uncertainty (Step 7) and monitoring and evaluating scenario impacts (Step 9). We found that although most studies mention uncertainty as an underlying rationale to employ PSP, uncertainty is often poorly evaluated. This may lead to misconceptions regarding the level of confidence with which results can be employed in assessments and decision-making processes. Uncertainty in PSP may arise from a variety of sources, including insufficient or erroneous data used to construct and test models, e.g., soil moisture and streamflow at high elevations (Wu et al. 2012, Capitani et al. 2016), problems in system understanding, e.g., the functional role of predator species in alpine ecosystems , or not having a full range of perspectives in the participatory workshops. This lack of diverse perspectives can then lead to insufficient representation of underlying processes, e.g., inadequate understanding of strategic national priorities for large-scale electrification (Roy et al. 2019) or low predictability of the system, e.g., random behavior in energy demand or nonlinear shifts from stable states due to eutrophication (Rockström et al. 2009. The MtSES research community can learn from the wider scientific community operating in other systems to improve the application of PSP in MtSES. The IPBES recommends that types, sources, and levels of uncertainty should always be critically evaluated and communicated regardless of the scenarios' goals, including the degree of confidence in the data and outputs, and the relative probabilities of diverse futures when statistically available. Similarly, the MtSES scientific community should set standards for best practices to provide robust and transparent evaluations of uncertainty and encourage research into new methods and their impact on decision making (Akçakaya et al. 2016, IPBES 2016. Future work should focus on monitoring whether scenario outcomes meet the model purpose and inform indicators for policy impact. It should also focus on evaluating the predictive capacity, learning, and feasibility of policies under different scenarios. Enhanced monitoring and evaluation of the https://www.ecologyandsociety.org/vol25/iss3/art6/ benefits of PSP could lead to a positive feedback in which more groups undertake PSP and experience its benefits ).

CONCLUSION
In this paper we help to frame future developments of MtSES research and application by providing the first systematic review on the state of PSP in MtSES. Results indicate that since 2006, there has been a steady rise in the application of PSP in MtSES, which has helped fill critical gaps in understanding data-scarce regions of the world. Our findings from MtSES indicate the potential for a substantial amount of learning to take place through the PSP process: by integrating diverse views, understanding local contexts, comparing trajectories of change in and across MtSES, highlighting complexities, uncertainties, and interdependencies, and potentially aiding adaptation to global environmental change. PSP can also be a useful tool to build cooperation and collaboration, shared understanding, and decision support. Overall, we found that PSP in MtSES is an inclusive, flexible, adaptable method that can be applied across scales, involving different actors and communication strategies. PSP can therefore help address some of the challenges, or paradoxes, of MtSES. However, because information about the "success" of PSP is not often documented explicitly in published studies, more research is needed to provide an evidence-based understanding of when and how such benefits are realized. Substantial progress has been made by using PSP to understand potential futures in MtSES, yet certain gaps remain. The geographic coverage of PSP in MtSES remains biased toward European countries, with less focus in Asian, African, and Latin American MtSES, particularly relative to their extent and human population. Transboundary collaborations are lacking, although many MtSES occur across administrative boundaries. PSP has tended to address certain themes, such as governance, economy, land use change, and biodiversity, but not others, such as gender equality, public health and sanitation, permafrost thaw, and the use and trade of medicinal resources, which are increasingly important for sustainable MtSES. The temporal scope of PSP is wide ranging, from near-term (14 y) to long-term (90 y) horizons, while analyses with nested temporal scales are less common. Though often stated as an intent, our review revealed low participation in terms of diversity and number of stakeholders at multiple stages of PSP, and with a limited set of institutions employing PSP.
To address these shortcomings and enhance effective future application, there is a need to improve the visibility and impact of PSP scholarship in national and international policy arenas, with a wider array of target audiences and systems of knowledge. These improvements could increase stakeholders' ability to take up research and to engage in agenda-setting. Two steps that require more attention in future work are (1) communicating uncertainty and (2) monitoring and evaluating scenario impact. We also encourage long-term capacity building in PSP, especially of early career researchers involving both South-North and South-South exchange to overcome biases in geographic coverage. This development will require substantive investments to expand and connect the MtSES research and practice communities, as well as a shift in scientific funding, regional cooperation, and science-policy frameworks. Achieving a greater degree of integration could involve analyzing systems over longer time periods, larger spatial extents, and expanding comparative research and subsequent assessment of benefits. This will require the MtSES research community to address the entrenched disadvantages of some countries to initiate research projects, and to reduce the logistical and financial burdens of studying across international boundaries. Future research may explore the extent to which PSP facilitated movement toward desired outcomes or decision making in MtSES (recognizing that this could take time to come to fruition) and which planning mechanisms or adaptive governance systems support scenario implementation most effectively. Results of this review may inform communities, organizations, companies, researchers, and governments tasked with anticipating a future without precedent for MtSES, and SES more broadly. August 13, 2017 4 ("scenario analy*" OR "scenario develop*" OR "scenario planning" OR "scenario") AND TOPIC: ("mountain*") Refined by: TOPIC: ("socio-ecological") 4 August 13, 2017 5 ("scenario analy*" OR "scenario develop*" OR "scenario planning" OR "scenario") AND TOPIC: ("mountain*") AND TOPIC: ("participat*model*") ORTOPIC: ("socioecological system") OR TOPIC: ("transdisciplinary" OR "interdisciplinary" OR "multidisciplinary") OR TITLE:("stakeholder") 130944 August 13, 2017 ! 6 ("scenario analy*" OR "scenario develop*" OR "scenario planning" OR "scenario") AND TOPIC: ("mountain*") AND TOPIC: ("participat*model*" OR "model" OR "socio-ecological") !